How Financial Motivation Can Be Measured in A Regression Model?

Stock Return Predictability: Is it There?
Andrew Ang
Columbia University and NBER
Geert Bekaert
Columbia University, NBER and CPER
We examine the predictive power of the dividend yields for forecasting excess returns,
cash flows, and interest rates. Dividend yields predict excess returns only at short
horizons together with the short rate and do not have any long-horizon predictive
power. At short horizons, the short rate strongly negatively predicts returns. These
results are robust in international data and are not due to lack of power. A present
value model that matches the data shows that discount rate and short rate movements
play a large role in explaining the variation in dividend yields. Finally, we find that
earnings yields significantly predict future cash flows. (JEL C12, C51, C52, E49,
F30, G12)
In a rational no-bubble model, the price-dividend ratio is the expected

value of future cash flows discounted with time-varying discount rates.
Because price-dividend ratios, or dividend yields, vary over time, dividend
yield variability can be attributed to the variation of expected cash flow
growth, expected future risk-free rates, or risk premia. The ‘‘conventional
wisdom’’ in the literature (see, among others, Campbell, 1991; Cochrane,
1992) is that aggregate dividend yields strongly predict excess returns, and
the predictability is stronger at longer horizons.1 Since dividend yields
only weakly predict dividend growth, conventional wisdom attributes
most of the variation of dividend yields to changing forecasts of expected
returns. We critically and comprehensively re-examine this conventional
wisdom regarding return predictability on the aggregate market.
We thank Xiaoyan Zhang for help with data. We thank Kobi Boudoukh, Michael Brandt, John
Campbell, John Cochrane, George Constantinides, Cam Harvey, Bob Hodrick, Charles Jones, Owen
Lamont, Martin Lettau, Jun Liu, Sydney Ludvigson, Chris Neely, Jim Poterba, Tano Santos, Bob
Shiller, Tim Simin, Ken Singleton, Rob Stambaugh, Meir Statman, Jim Stock, Ane Tamayo, Samuel
Thomson, Ivo Welch, Jeff Wurgler, and seminar participants at Columbia University, Duke University,
INSEAD, London Business School, the NY Federal Reserve, NYU, Stanford University, UC San Diego,
Wharton, Yale, the American Finance Association, an NBER Asset Pricing meeting, the European
Finance Association, and the Western Finance Association for helpful comments. We also thank three
anonymous referees and Yacine Aı̈t-Sahalia (the editor) for valuable, extensive comments that greatly
improved the article. Andrew Ang thanks the Chazen Institute at Columbia Business School for financial
support. Geert Bekaert thanks the NSF for financial support. Address correspondence to Geert Bekaert,
Columbia Business school, Room 802, Uris Hall, 3022 Broadway, New York, NY 10027, or e-mail:
gb241@columbia.edu.
1
Among those examining the predictive power of the dividend yield on excess stock returns are Fama and
French (1988), Campbell and Shiller (1988a,b), Goetzmann and Jorion (1993, 1995), Hodrick (1992),
Stambaugh (1999), Wolf (2000), Goyal and Welch (2003, 2004), Engstrom (2003), Valkanov (2003),
Lewellen (2004), and Campbell and Yogo (2006).
The Author 2006. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights
reserved. For permissions, please email: journals.permissions@oxfordjournals.org.
doi:10.1093/rfs/hhl021 Advance Access publication July 6, 2006
The Review of Financial Studies / v 20 n 3 2007
Our main findings can be summarized as follows. First, the statistical

inference at long horizons critically depends on the choice of standard
errors. With the standard Hansen–Hodrick (1980) or Newey–West (1987)
standard errors, there is some evidence for long horizon predictability,
but it disappears when we correct for heteroskedasticity and remove the
moving average structure in the error terms induced by summing returns
over long horizons (Richardson and Smith 1991, Hodrick 1992,
Boudoukh and Richardson, 1993).
Second, we find that the most robust predictive variable for future
excess returns is the short rate, but it is significant only at short horizons.2
Whereas, the dividend yield does not univariately predict excess returns,
the predictive ability of the dividend yield is considerably enhanced, at
short horizons, in a bivariate regression with the short rate. To mitigate
data snooping concerns (Lo and MacKinlay 1990, Bossaerts and Hillion
1999, Ferson, Sarkissian, and Simin 2003, Goyal and Welch 2004), we
confirm and strengthen this evidence using three other countries: the
United Kingdom, France, and Germany.
Third, the dividend yield’s predictive power to forecast future dividend
growth is not robust across sample periods or countries. We find that high
dividend yields are associated with high future interest rates. While the
statistical evidence for interest rate predictability is weak, the same posi-
tive relationship is implied by an economic model, and we observe the
same patterns across countries.
To help interpret our findings and to deepen our understanding of the
data, we provide additional economic analysis. First, we build a nonlinear
present value model with stochastic discount rates, short rates, and divi-
dend growth, that matches our evidence on excess return predictability.
As is true in the data, the model implies that the dividend yield only
weakly predicts future cash flows but is positively related to future move-
ments in interest rates. While excess discount rates still dominate the
variation in price-dividend ratios, accounting for 61%, of the variation
short rate movements account for up to 22% of the variation. In compar-
ison, dividend growth accounts for around 7% of the variance of price-
dividend ratios. The rest of the variation is accounted for by covariance
terms.
Because many studies, particularly in the portfolio choice literature, use
univariate dividend yield regressions to compute expected returns
(Campbell and Viceira, 1999), we use the nonlinear present value model
to examine the fit of regression-based expected returns with true expected
returns. Consistent with the data, we find that a univariate dividend yield
2
Authors examining the predictability of excess stock returns by the nominal interest rate include Fama
and Schwert (1977), Campbell (1987), Breen, Glosten, and Jagannathan (1989), Shiller and Beltratti
(1992), and Lee (1992).
652
Stock Return Predictability
regression provides a rather poor proxy to true expected returns. How-

ever, using both the short rate and dividend yield considerably improves
the fit, especially at short horizons.
Second, using the present value model, we show that long-horizon
statistical inference with the standard Hansen–Hodrick (1980) or
Newey–West (1987) standard errors is treacherous. We find that both
Hansen–Hodrick and Newey–West standard errors lead to severe over-
rejections of the null hypothesis of no predictability at long horizons but
that standard errors developed by Hodrick (1992) retain the correct size in
small samples. The power of Hodrick t-statistics exceeds 0.60 for a 5%
test for our longest sample. Moreover, when we pool data across differ-
ent countries, the power for our shortest sample increases to 74%. Hence,
lack of power is unlikely to explain our results.3
Finally, we focus on expanding the information set to obtain a poten-
tially better estimate of true value-relevant cash flows in the future.
Dividends may be potentially poor instruments because dividends are
often manipulated or smoothed. Bansal and Lundblad (2002) and Bansal
and Yaron (2004) argue that dividend growth itself follows an intricate
ARMA process. Consequently, it is conceivable that more than one
factor drives the dynamics of cash flows. One obvious way to increase
the information set is to use earnings. Lamont (1998) argues that the
earnings yield has independent forecasting power for excess stock returns
in addition to the dividend yield. When we examine the predictive power
of the earnings yield for both returns and cash flows, we find only weak
evidence for Lamont’s excess return predictability results. However, we
detect significant predictability of future cash flows by earnings yields.
This article is organized as follows. Section 1 describes the data. Section 2
contains the main predictability results for returns, while Section 3 discusses
cash flow and interest rate predictability by the dividend yield. Section 4
develops a present value model under the null and various alternative
models to interpret the empirical results. In Section 5, we conduct a size
and power analysis of Hodrick (1992) standard errors. Section 6 investi-
gates the predictive power of the earnings yield for excess returns and cash
flows. Section 7 concludes and briefly discusses a number of contempora-
neous papers on stock return predictability. It appears that the literature is
converging to a new consensus, substantially different from the old view.
3
Given the excellent performance of Hodrick (1992) standard errors, we do not rely on the alternative
inference techniques that use unit root, or local-to-unity, data generating processes (see, among others,
Richardson and Stock 1989, Richardson and Smith 1991, Elliot and Stock 1994, Lewellen 2004, Torous,
Valkanov, and Yan 2004, Campbell and Yogo 2006, Polk, Thompson, and Vuolteenaho 2006, Jansson
and Moreira 2006). One major advantage of Hodrick standard errors is that the set up can handle
multiple regressors, whereas the inference with unit root type processes relies almost exclusively on
univariate regressors. The tests for multivariate predictive regressions using local-to-unity data generating
processes developed by Polk et al. (2006) involve computationally intensive bootstrapping procedures.
This test also has very poor size properties under the nonlinear present value we present in Section 4.
These results are available upon request.
653
1. Data
We work with two data sets, a long data set for the United States, United
Kingdom and Germany and a shorter data set for a sample of four
countries (United States, United Kingdom, France, and Germany). In
the data, dividend and earnings yields are constructed using dividends
and earnings summed up over the past year. Monthly or quarterly fre-
quency dividends and earnings are impossible to use because they are
dominated by seasonal components.
We construct dividend growth and earnings growth from these ratios,
producing rates of annual dividend or earnings growth over the course of
a month or a quarter. To illustrate this construction, suppose we take the
frequency of our data to be quarterly. We denote log dividend growth at a
quarterly frequency as gtd,4 , with the superscript 4 to denote that it is
constructed using dividends summed up over the past year (four quar-
ters). We compute gtd,4 from dividend yields D4t =Pt , where the dividends
are summed over the past year, using the relation

D4t =Pt Pt
gtd,4 ¼ log , ð1Þ
D4t1 =Pt1 Pt1
where D4t ¼ Dt þ Dt1 þ Dt2 þ Dt3 represents dividends summed over

the past year and Pt =Pt1 is the price return over the past quarter.
In our data, the long sample is at a quarterly frequency and the short
sample is at a monthly frequency. In the case of the monthly frequency,
we append dividend yields, earnings yields, dividend growth and earnings
growth with a superscript of 12 to indicate that dividends and earnings
have been summed over the past 12 months. We also denote log dividend
yields by lower case letters. Hence dy4t ¼ logðD4t =Pt Þ, in the case of
quarterly data and dy12 12
t ¼ logðDt =Pt Þ in the case of monthly data. We
also use similar definitions for log earnings yields: ey4t and ey12
t .
1.1 Long sample data

Our US data consists of price return (capital gain only), total return
(capital gain plus dividend), and dividend and earnings yields on the
Standard & Poor’s Composite Index from June 1935 to December 2001.
This data is obtained from the Security Price Index Record, published by
Standard & Poor’s Statistical Service. Lamont (1998) uses the same data
set over a shorter period. The long-sample UK data comprises price
returns and total returns on the Financial Times (FT) Actuaries Index,
and we construct implied dividend yields from these series. For our
German data, we take price returns, total returns, and dividend yields
on the composite DAX (CDAX) index from the Deutsche Borsche. The
long-sample UK and Ge-rman data span June 1953 to December 2001
654
and were purchased from Global Financial Data. All the long sample
data for the United States, United Kingdom, and Germany are at the
quarterly frequency, and we consequently use three-month T-bills as
quarterly short rates.
Panel A of Table 1 lists summary statistics. US earnings growth is
almost as variable as returns, whereas the volatility of dividend growth
is less than half the return volatility. The variability of UK and German
dividend growth rates is of the same order of magnitude as that of
returns. The instruments (short rates, dividend and earnings yields) are
all highly persistent. Because the persistence of these instruments plays a
crucial role in the finite sample performance of predictability test statis-
tics, we report test statistics under the null of a unit root and a stationary
process in Panel A. Investigating both null hypotheses is important
because unit root tests have very low power to reject the null of a
stationary, but persistent, process.
In the United Kingdom and Germany, dividend yields are unambigu-
ously stationary, as we reject the null of a unit root and fail to reject the
null of stationarity at the 5% level. For the US dividend yield, the
evidence for non-stationarity is weak as we fail to reject either hypothesis.
This is surprising because the trend toward low dividend yields in the
1990s has received much attention. Figure 1 plots dividend yields for the
Table 1
Sample moments, unit root, and stationarity tests
Excess Short Dividend Earnings Dividend Earnings

return rate yield yield growth growth
Panel A: Long-sample data

US S&P Data, June 1935–December 2001
Mean 0.0749 0.0409 0.0403 0.0768 0.0532 0.0548
Stdev 0.1684 0.0317 0.0150 0.0297 0.0658 0.1572
Auto 0.1173 0.9548 0.9504 0.9517 0.4071 0.3832
Test statistics
H0: unit root 14.50** 2.194 1.187 1.183 10.83** 10.55**
H0: stationary 0.073 0.635* 0.372 0.336 0.035 0.026
UK FT Data, June 1953–December 2001
Mean 0.0563 0.0751 0.0478 0.0670
Stdev 0.1938 0.0331 0.0131 0.1866
Auto 0.0907 0.9400 0.8290 0.0486
Test statistics
H0: unit root 12.66** 2.559 4.125** 14.64**
H0: stationary 0.037 0.637* 0.199 0.068
Germany DAX Data, June 1953–December 2001
Mean 0.0577 0.0467 0.0287 0.0788
Stdev 0.1921 0.0198 0.0090 0.2086
Auto 0.0851 0.9376 0.9087 0.1136
Test statistics
H0: unit root 12.89** 3.036* 3.336* 12.34**
H0: stationary 0.091 0.313 0.328 0.156
655
Table 1
(continued)
Excess Short Dividend Earnings Dividend Earnings

return rate yield yield growth growth
Correlations of excess returns, June 1953–December 2001

United States United Kingdom
United Kingdom 0.6281
Germany 0.5118 0.4598
Panel B: MSCI data

February 1975–December 2001
United States
Mean 0.0576 0.0745 0.0353 0.0744 0.0529 0.0501
Stdev 0.1513 0.0341 0.0143 0.0305 0.0589 0.0883
Auto 0.0044 0.9675 0.9892 0.9867 0.3187 0.1490
United Kingdom
Mean 0.0604 0.0989 0.0456 0.0874 0.0812 0.0651
Stdev 0.1824 0.0355 0.0123 0.0346 0.0707 0.0973
Auto 0.0184 0.9615 0.9701 0.9758 0.0524 0.2346
France
Mean 0.0542 0.0906 0.0415 0.0675 0.0782 0.0690
Stdev 0.2079 0.0476 0.0188 0.0397 0.0849 0.5868
Auto 0.0745 0.8741 0.9849 0.9627 0.0068 0.0765
Germany
Mean 0.0498 0.0563 0.0359 0.0688 0.0643 0.0564
Stdev 0.1925 0.0241 0.0125 0.0287 0.0876 0.2217
Auto 0.0665 0.9764 0.9860 0.9836 0.0936 0.1781
Correlations of excess returns

United States United Kingdom France
United Kingdom 0.5960
France 0.5237 0.5184
Germany 0.4951 0.4742 0.6178
Panel A reports summary statistics of long-sample data for the United States, United Kingdom, and
Germany, all at a quarterly frequency. Panel B reports statistics for monthly frequency Morgan Stanley
Capital International MSCI data. Excess returns and short rates are continuously compounded. Sample
means and standard deviations (Stdev) for excess
pffiffiffiffi pffiffiffiffiffi returns, dividend, and earnings growth have been
annualized by multiplying by 4 (12) and 4 ð 12Þ, respectively, for the case of quarterly (monthly)
frequency data. Short rates for the long-sample (MSCI) data are three-month T-bill returns (one month
EURO rates). Dividend and earnings yields, and the corresponding dividend and earnings growth are
computed using dividends or earnings summed up over the past year. In Panel A, the unit root test is the
Phillips and Perron (1988) test for the estimated regression xt ¼ þ xt1 þ ut under the null
xt ¼ xt1 þ ut . The critical values corresponding to p-values of 0.01, 0.025, 0.05, and 0.10 are 3:46,
3:14, 2:88, and 2:57, respectively. The test for stationarity is the Kwiatkowski et al. (1992) test. The
critical values corresponding to p-values of 0.01, 0.025, 0.05, and 0.10 are 0.739, 0.574, 0.463, 0.347,
respectively.
*p<0.05.
**p<0.01.
United States, United Kingdom, and Germany. For the United

Kingdom, the dividend yield also declined during the late 1990s, but the
United Kingdom experienced similar low level dividend yields during the
late 1960s and early 1970s. For Germany, there is absolutely no trend in
the dividend yield. If a time trend in dividend yields is a concern for
656
0.12
US
UK
Germany
0.1
0.08
0.06
0.04
0.02
0
1940 1950 1960 1970 1980 1990 2000
Figure 1
Dividend yields over the long sample
We plot dividend yields from June 1935 to December 2001 for the United States and from March 1953 to
December 2001 for the United Kingdom and Germany.
interpreting the evidence on excess return predictability using the divi-

dend yield, international data are clearly helpful. Present value models
which impose transversality also imply that dividend yields must be
stationary.
Interest rates are also highly persistent variables. While German inter-
est rates appear to be stationary, there is some evidence of borderline non-
stationary behavior for both US and UK interest rates. Most economic
models also imply that interest rates are stationary (Clarida, Galı́, and
Gertler, 1999). Our present value model incorporates realistic persistence
in short rates, but because of the high persistence of the short rate, we
check the robustness of interest rate predictability by using a detrended
short rate.
1.2 Short sample MSCI data

The data for the United States, United Kingdom, France, and Germany
consist of monthly frequency price indices (capital appreciation only),
total return indices (including income), and valuation ratios from Morgan
Stanley Capital International (MSCI) in local currency, from February
657
1975 to December 2001. We use the one-month EURO rate from Data-
stream as the short rate.
Panel B of Table 1 shows that the United States has the least variable
stock returns with the least variable cash flow growth rates. The extreme
variability of French earnings growth rates is primarily due to a few
outliers between May 1983 and May 1984, when there are very large
movements in price-earnings ratios. Without these outliers, the French
earnings growth variability drops to 33%. The variability of short rates,
dividend, and earnings yields is similar across countries. The equity pre-
mium for the United States, France, Germany, and the United Kingdom
roughly lies between 4 and 6% during this sample period. Dividend yields
and short rates are again very persistent over the post-1975 sample. We
also report excess return correlations showing that correlations range
between 0.47 and 0.60. The correlations for the United States, United
Kingdom, and Germany are similar to the correlations over the post-1953
period reported in Panel A.
2. The Predictability of Equity Returns
2.1 Predictability regressions

Denote the gross return on equity by Ytþ1 ¼ ðPtþ1 þ Dtþ1 Þ=Pt and the
continuously compounded return by ytþ1 ¼ logðYtþ1 Þ. The main regres-
sion we consider is
ytþk ¼ k þ k zt þ "tþk,k ,

~ ð2Þ
where ~ ytþk ¼ ð=kÞ½ðytþ1 rt Þ þ ::: þ ðytþk rtþk1 Þ is the annualized

k-period excess return for the aggregate stock market, rt is the risk-free
rate from t to t þ 1, and ytþ1 rt is the excess one period return from
time t to t þ 1. A period is either a month ð ¼ 12Þ or a quarter ð ¼ 4Þ.
All returns are continuously compounded. The error term "tþk,k follows a
MAðk 1Þ process under the null of no predictability ðk ¼ 0Þ because
of overlapping observations. We use log dividend yields and annualized
continuously compounder risk-free rates as instruments in zt .
We estimate the regression (2) by OLS and compute standard errors of
the parameters ¼ ð k Þ following Hodrick (1992). Using generalized
pffiffiffiffi
method of moments, (GMM) has an asymptotic distribution T ð^ Þ
a
Nð0,Þ where ¼ Z01 S0 Z01 , Z0 ¼ Eðxt xt Þ, and xt ¼ ð1 zt Þ . Hodrick
exploits covariance stationarity to remove the overlapping nature of the
error terms in the standard error computation. Instead of summing "tþk,k
into the future to obtain an estimate of S0 , Hodrick sums xt xtj into the
past and estimates S0 by
658
XT
^0 ¼ 1
S wkt wkt , ð3Þ
T t¼k
where
!
X
k1
wkt ¼ "tþ1,1 xti :
i¼0
We show in section 5 that the performance of Hodrick (1992) standard

errors is far superior to the Newey–West (1987) standard errors or the
robust GMM generalization of Hansen and Hodrick (1980) standard
errors (see Appendix A) typically run in the literature. Hence, our pre-
dictability evidence exclusively focuses on Hodrick t-statistics. Mindful of
Richardson’s (1993) critique of focusing predictability tests on only one
particular horizon k, we also compute joint tests across horizons. For the
quarterly (monthly) frequency data, we test for predictability jointly
across horizons of 1, 4, and 20 quarters (1, 12, and 60 months). Appendix
B details the construction of joint tests across horizons accommodating
Hodrick standard errors. Finally, when considering predictability in
multiple countries, we estimate pooled coefficients across countries and
provide joint tests of the null of no predictability. Pooled estimations
mitigate the data-mining problem plaguing US data and, under the null of
no predictability, enhance efficiency because the correlations of returns
across countries are not very high (Table 1). Appendix C details the
econometrics underlying the pooled estimations.
2.2 Return predictability in the United States

We report results for several sample periods, in addition to the full sample
1935–2001. Interest rate data are hard to interpret before the 1951 Treasury
Accord, as the Federal Reserve pegged interest rates during the 1930s and
the 1940s. Hence, we examine the post-Accord period, starting in 1952.
Second, the majority of studies establishing strong evidence of predictabil-
ity use data before or up to the early 1990s. Studies by Lettau and
Ludvigson (2001) and Goyal and Welch (2003) point out that predictability
by the dividend yield is not robust to the addition of the 1990s decade.
Hence, we separately consider the effect of adding the 1990s to the sample.
We start by focusing on a univariate regression with the dividend yield
as the regressor. Figure 2 shows the slope coefficients for three different
sample periods, using the quarterly US S&P data. The left-hand column
reports the dividend yield coefficients, whereas the right-hand column
reports t-statistics computed using Newey–West (1987), robust
Hansen–Hodrick (1980), and Hodrick (1992) standard errors. For the
659
1935-2001 Dividend Yield Coefficient 1935-2001 T-statistic

0.115 5.5
Robust Hansen−Hodrick
Hodrick
5 Newey−West
0.11
4.5
0.105 4
3.5
0.1
Coefficient
T−stat
2.5
0.095
0.09 1.5
1
0.085
0.5
0.08 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Horizon (quarters) Horizon (quarters)

0.12 5.5
Robust Hansen−Hodrick
Hodrick
5 Newey−West
0.11
4.5
0.1
4
0.09 3.5
Coefficient
3
T−stat
0.08
2.5
0.07 2
1.5
0.06
1
0.05
0.5
0.04 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20

0.25 5.5
5
0.24
4.5
0.23
4
0.22
3.5
Coefficient
0.21 3
T−stat
0.2 2.5
2
0.19
1.5
0.18
1 Robust Hansen−Hodrick
Hodrick
0.17 Newey−West
0.5
0.16 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Figure 2
Dividend yield coefficients and t-statistics from US regressions
The left (right) column shows the dividend yield coefficients k (t-statistics) in the regression
~ytþk ¼ þ k dy4t þ "tþk,k , where ~ytþk is the cumulated and annualized k-quarter ahead excess return
and dy4t is the log dividend yield. T-statistics are computed using Robust Hansen-Hodrick (1980),
Hodrick (1992) or Newey-West (1987) standard errors. The quarterly data is from Standard and Poors.
Newey–West errors, we use k þ 1 lags. The coefficient pattern is similar

across the three periods, but the coefficients are twice as large for the
period omitting the 1990s from the sample. For the other two periods
(1935–2001 and 1952–2001), the one-period coefficient is about 0.110,
660
rises until the one-year horizon, and then decreases, before increasing
again near 20 quarters.
Over 1935–2001, the Hodrick t-statistic is above 2 only for horizons 2–
4 quarters. However, there is no evidence of short-run predictability (at
the one-quarter horizon) or long-horizon predictability. We draw a very
different picture of predictability if we use Newey–West or robust
Hansen–Hodrick t-statistics, which are almost uniformly higher than
Hodrick t-statistics. Using Newey–West standard errors, the evidence in
favor of predictability would extend to eight quarters for the full sample.
Over the 1952–2001 sample, there is no evidence of predictability,
whereas for the 1935–1990 period, the evidence for predictability is very
strong, whatever the horizon, with all three t-statistics being above 2.4.
Table 2 summarizes the excess return predictability results for horizons
of one month (quarter), one year, and five years. We only report
t-statistics using Hodrick standard errors. In addition to the sample
periods shown in Figure 2, we also show the 1952–1990 period, which is
close to the 1947–1994 sample period in Lamont (1998). When we omit
the 1990s, we confirm the standard results found by Campbell and Shiller
(1988a,b) and others: the dividend yield is a significant predictor of excess
Table 2
Predictability of US excess returns
Univariate
regression Bivariate regression
12
k-mths dy r dy12 2 test
Panel A: quarterly S&P data

1935–2001 1 0.1028 1.0888 0.0857 0.070
(1.824) (1.608) (1.530)
4 0.1128 0.5596 0.1032 0.104
(2.030)* (0.827) (1.865)
20 0.1028 0.3187 0.0952 0.323
(1.364) (0.598) (1.255)
1952–2001 1 0.0979 2.1623 0.1362 0.003**
(1.541) (2.912)** (2.152)*
4 0.1060 1.4433 0.1313 0.041*
(1.546) (1.930) (1.921)
20 0.0594 0.4829 0.0774 0.714
(0.477) (0.745) (0.600)
1935–1990 1 0.2203 1.0380 0.1917 0.027*
(2.416)* (1.543) (2.126)*
4 0.2383 0.4865 0.2254 0.008**
(3.097)** (0.714) (3.006)**
20 0.1787 0.3229 0.1719 0.017*
(2.819)** (0.569) (2.832)**
1952–1990 1 0.2962 2.7329 0.4125 0.002**
(2.783)** (3.504)** (3.672)**
4 0.3070 1.9840 0.3935 0.000**
(3.000)** (2.508)* (3.700)**
20 0.1689 0.7120 0.2057 0.052
(1.916) (1.087) (2.387)*
661
Table 2
(continued)
Univariate
k-mths dy12 r dy12 2 test
Panel B: monthly MSCI data

1975–2001 1 0.0274 2.4358 0.1364 0.057
(0.405) (2.388)* (1.626)
12 0.0106 1.2470 0.0669 0.395
(0.141) (1.361) (0.744)
60 0.0884 0.3451 0.1207 0.857
(0.475) (0.238) (0.397)
We estimate regressions of the form ~ytþk ¼ k þ zt þ tþk,k where y

~tþk is the cumulated and annualized
k-period ahead excess return, with instruments zt being log dividend yields or risk-free rates and log
dividend yields together. T-statistics in parentheses are computed using Hodrick (1992) standard errors.
For Panel A (B), horizons k are quarterly (monthly). The 2 test column reports a p-value for a test that
both the risk-free rate and log dividend yield coefficients are jointly equal to zero.
*p<0.05.
**p<0.01.
returns at all horizons. However, when we use all the data, we only find
5% significance at the one-year horizon for the longest sample. A test for
predictability by the dividend yield jointly across horizons rejects with a
p-value of 0.014 for the 1935–1990 period, but fails to reject with a
p-value of 0.587 over the whole sample, even though the shorter horizon
t-statistics all exceed 1.5 in absolute value. While it is tempting to blame
the bull market of the 1990s for the results, our data extend until the end
of 2001 and hence incorporate a part of the bear market that followed.
For the 1975–2001 sample, reported in Panel B, the dividend yield also
fails to predict excess returns.
Table 2 also reports bivariate regression results with the short rate as an
additional regressor. For the post-Treasury Accord 1952–2001 sample, a
1% increase in the annualized short rate decreases the equity premium by
about 2.16%. The effect is significant at the 1% level. A joint test on the
interest rate coefficients across horizons rejects strongly for both the
1952–2001 period ðp-value ¼ 0:004Þ and the 1952–1990 period
ðp-value ¼ 0:000Þ. The predictive power of the short rate dissipates
quickly for longer horizons but remains borderline significant at the 5%
level at the one-year horizon.4 If expected excess returns are related only
to short rates and short rates follow a univariate autoregressive process,
the persistence of the interest rate (0.955 in Table 1) implies that the
coefficient on the short rate should tend to zero slowly for long horizons.
In fact, the decay rate should be 1=k ð1 k Þ=ð1 Þ for horizon k. The
4
The results do not change when a detrended short rate is used instead of the level of the short rate or when
we use a dummy variable over the period from October 1979 to October 1982 to account for the monetary
targeting period.
662
decay rate in data is clearly more rapid, indicating that either expected
excess returns or risk-free rates, or both, are multifactor processes.
In the bivariate regression, the dividend yield coefficient is only signifi-
cant at the 5% level for the one-quarter horizon. Joint tests reject at the 1%
(5%) level for the one-quarter (four-quarter) horizon but fail to reject at
long horizons. When we omit the 1990s, the predictive power of the short
rate becomes even stronger. A joint predictability test still fails to reject the
null of no predictability at long horizons, but the p-value is borderline
significant (0.052). Over the 1975–2001 sample, the coefficient on the short
rate remains remarkably robust and is significant at the 5% level. While the
coefficient on the dividend yield is no longer significantly different from
zero, it is similar in magnitude to the full sample coefficient and a joint test
is borderline significant ðp-value ¼ 0:057Þ. The Richardson (1993) joint
predictability test over all horizons and both predictors rejects at the 1%
level in the samples excluding the 1990s and the full sample, rejects at the
5% level for 1952–2001, and rejects at the 10% level for 1975–2001.
Looking at the 1951–2001 and 1975–2001 samples, the evidence for the
bivariate regression at short horizons is remarkably consistent. Moreover,
the coefficient on the dividend yield is larger in the bivariate regression
than in the univariate regression. This suggests that the univariate regres-
sion suffers from an omitted variable bias that lowers the marginal impact
of dividend yields on expected excess returns. Engstrom (2003), Menzly,
Santos, and Veronesi (2004), and Lettau and Ludvigson (2005) also note
that a univariate dividend yield regression may understate the dividend
yield’s ability to forecast returns.
2.3 Predictability of excess returns in four countries

The weak predictive power of the univariate dividend yield in the full
sample may simply be a small sample phenomenon due to the very special
nature of the last decade for the US stock market. Alternatively, the
conventional wisdom of strong long-horizon excess return predictability
by dividend yields before 1990 may be a statistical fluke. International
evidence can help us to sort out these two interpretations of the data and
check the robustness of predictability patterns observed in US data.
Figure 3 displays the univariate dividend yield coefficients and their
t-statistics using Hodrick standard errors in the 1975–2001 sample. First,
none of the patterns in other countries resembles the US pattern. For
France and Germany, and to a lesser degree for the United Kingdom, the
coefficients first increase with horizon, then decrease, and finally increase
again. This is roughly the pattern we see in US data for the longer
samples. However, for France and Germany, the coefficients are small
at short horizons and are negative for many horizons. They are also
never statistically significant. The UK coefficient is larger and remains
positive across horizons: it is also significantly different from zero at the
663
US dy coefficient US t−stat
0.1
2
0.05
1
0
−0.05 0
−0.1 −1
UK dy coefficient UK t−stat
0.35
2
0.3
0.25 1
0.2
0
0.15
0.1 −1
FR dy coefficient FR t−stat
0.06
2
0.04
1
0.02
0 0
−0.02 −1
GR dy coefficient GR t−stat
0.04
2
0.02
1
0
−0.02 0
−0.04 −1
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Horizon (months) Horizon (months)
Figure 3
Dividend yield coefficients and t-statistics in four countries
The left (right) column shows the dividend yield coefficients k (t-statistics) in the regression
~ytþk ¼ þ k dy12
t þ "tþk,k , where ~ytþk is the cumulated and annualized k-month ahead excess return
and dy12
t is the log dividend yield. T-statistics are computed using Hodrick (1992) standard errors. The
monthly data is from MSCI and the sample period is from 1975 to 2001.
very shortest horizons. These results are opposite to the results in a recent
study by Campbell (2003), who reports strong long-horizon predictability
for France, Germany, and the United Kingdom over similar sample
periods. We find that Campbell’s conclusions derive from the use of
Newey–West (1987) standard errors, and the predictability disappears
when Hodrick (1992) standard errors are employed.
For the United Kingdom and Germany, we also investigate the longer
1953–2001 sample in Panel A of Table 3. The first column reports the
univariate dividend yield coefficients. We only find significance at the
one-year horizon for the UK, but the coefficients are all positive and
more than twice as large as the US coefficients.5 Germany’s dividend yield
5
For the United Kingdom, we also looked at a sample spanning 1935–2001, where we find a significant
univariate dividend yield coefficient at the five-year horizon but not at the one-quarter horizon.
664
Table 3
Excess return regressions across countries
Univariate
k-qtrs dy4 only J-test r dy4 2 Test J-test
Panel A: quarterly long-sample data

United Kingdom, 1953–2001 1 0.2280 1.2148 0.2596 0.069
(1.449) (1.500) (1.675)
4 0.2600 0.7138 0.2777 0.041*
(2.075)* (0.896) (2.273)*
20 0.1290 0.1075 0.1264 0.326
(1.403) (0.182) (1.338)
Germany, 1953–2001 1 0.0740 3.4079 0.1083 0.030*
(0.827) (2.545)* (1.204)
4 0.1235 2.1027 0.1443 0.094
(1.366) (1.679) (1.610)
20 0.0415 0.4758 0.0372 0.814
(0.397) (0.489) (0.360)
Pooled United States, 1 0.1230 0.496 1.958 0.1600 0.001** 0.133
United Kingdom, (1.964)* (2.927)** (2.626)**
Germany, 1953–2001 4 0.1523 0.344 1.2561 0.1754 0.008** 0.307
(2.268)* (1.909) (2.709)**
20 0.0657 0.495 0.0157 0.0653 0.637 0.658
(0.763) (0.0319) (0.818)
Univariate
k-mths dy12 only J-test r dy12 chi 2 test J-test
Panel B: monthly MSCI data

Pooled United States, 1 0.0560 0.096 1.8161 0.1640 0.031* 0.016*
United Kingdom, Germany, (0.866) (2.718)** (2.222)*
France, 1975–2001 12 0.0386 0.327 1.1392 0.1060 0.229 0.113
(0.533) (2.045)* (1.337)
60 0.0169 0.663 0.1799 0.0035 0.699 0.887
(0.130) (0.405) (0.033)
We estimate regressions of the form ~ytþk ¼ k þ zt þ "tþk,k where y~tþk is the cumulated and annualized
k-period ahead excess return, with instruments zt being log dividend yields or risk-free rates and log
dividend yields together. T-statistics in parentheses are computed using Hodrick (1992) standard errors.
Panel A estimates the regression pooling data across the United States, United Kingdom, and Germany
on data from 1953–2001. The estimates listed in the United Kingdom and Germany panels allow each
country to have its own predictive coefficients and intercepts, but we compute Seemingly Unrelated
Regression (SUR) standard errors following the method outlined in the Appendix. The coefficients listed
in the pooled panel are produced by constraining the predictive coefficients to be the same across
countries. In Panel B, monthly frequency MSCI data is used from 1975–2001. The column labeled ‘‘2
test’’ reports a p-value for a test that both the risk-free rate and log dividend yield coefficients are jointly
equal to zero. The ‘‘J-test’’ columns report p-values for a 2 test of the overidentifying restrictions.
*p<0.05.
**p<0.01.
coefficients are the same order of magnitude than those of the United
States, but are all insignificant.
Figure 4 displays the coefficient patterns for the annualized short rate
and its associated t-statistics in the bivariate regression for the 1975–2001
sample. Strikingly, this coefficient pattern is very robust across countries.
For all countries, the one-month coefficient is negative, below 3 for
665
US rf coefficient US t−stat
1 2
1
0
0
−1
−1
−2
−2
−3 −3
UK rf coefficient UK t−stat
−0.5 2
1
−1
0
−1.5
−1
−2
−2
−2.5 −3
FR rf coefficient FR t−stat
0.5 2
0 1
−0.5 0
−1 −1
−1.5 −2
−2 −3
GR rf coefficient GR t−stat
2 2
1
0
0
−1
−2
−2
−4 −3
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Horizon (months) Horizon (months)
Figure 4
Short rate coefficients from bivariate regressions in four countries
The left (right) column shows the risk-free rate coefficients k (t-statistics) from the bivariate regression
~ytþk ¼ þ zt k þ "tþk,k , where ~ytþk is the cumulated and annualized k-month ahead excess return and
zt ¼ ðrt dy12
t Þ contains the annualized risk-free rate and the log dividend yield. We report only the short
rate coefficient. T-statistics are computed using Hodrick (1992) standard errors. The monthly data is
from MSCI and the sample period is from 1975 to 2001.
Germany and around 1:5 for France. The coefficient monotonically

increases with the horizon, leveling off around 0.35 for the United States,
0.13 for France, 1.41 for Germany, and 0:74 for the United Kingdom.
The t-statistics are larger in absolute magnitude for short horizons. In
particular, at the one-month horizon, the short rate coefficients are
statistically different from zero for the United States and Germany and
the t-statistics are near 1.5 (in absolute value) for the United Kingdom
and France.
Panel A of Table 3 reports the bivariate coefficients for the long sample
for the United Kingdom and Germany. Both countries have negative
coefficients on the short rate. For Germany, the short rate coefficient is
highly significant, while the UK t-statistic is only 1:5. For both coun-
tries, the short rate coefficients increase with horizon and turn positive at
666
the five-year horizon. Similar to the United States, the dividend yield
coefficients are larger in the bivariate regressions than in the univariate
regression. However, the dividend yield coefficient is still only signifi-
cantly different from zero at the 5% level in the United Kingdom at the
one-year horizon.
To obtain more clear-cut conclusions, Table 3 reports pooled predict-
ability coefficients and tests. We pool across the United States, United
Kingdom, and Germany for the long sample in Panel A, and pool across
all four countries in Panel B. The univariate dividend yield regression
delivers mixed evidence across the two samples. For the long sample, the
dividend yield coefficients are larger than 0.10 at the one-quarter and one-
year horizons and statistically significant at the 5% level. The joint pre-
dictability tests for the shorter sample reveal a pattern of small dividend
yield coefficients that decrease with horizon and are never significantly
different from zero. We also report a J-test of the overidentifying restric-
tions for the joint estimation (see Appendix C). This test fails to reject for
all horizons in both the long and short samples, which suggests that
pooling is appropriate.
What is most striking about the bivariate regression results across the
long and short samples is the consistency of the results. At the one-period
forecasting horizon, the short rate coefficient is 1:96 in the long sample
and 1:82 in the short sample, both significant at the 1% level. The
bivariate regression also produces a dividend yield coefficient around
0.16 that is significant at the 1% (5%) level in the long (short) sample.
Not surprisingly, the joint test rejects at the 5% level. However, for the
short sample, the test of the overidentifying restrictions rejects at the 5%
level, suggesting that pooling may not be appropriate for this horizon.
For longer horizons, this test does not reject, and the evidence for
predictability weakens. Nevertheless, for the long sample, we still reject
the null of no predictability at the 1% level for the one-year horizon.
We conclude that whereas the dividend yield is a poor predictor of
future returns in univariate regressions, there is strong evidence of pre-
dictability at short horizons using both dividend yields and short rates as
instruments. The short rate is the stronger predictor and predicts excess
returns with a coefficient that is negative in all four countries that we
consider.
3. Do Dividend Yields Predict Cash flows or Interest Rates?

Our predictability results overturn some conventional, well-accepted
results regarding the predictive power of dividend yields for stock returns.
The dividend yield is nonetheless a natural predictor for stock returns.
Define the discount rate t as the log conditional expected total return,
lnðEt ½Ytþ1 Þ:
667
expð t Þ ¼ Et ½ðPtþ1 þ Dtþ1 Þ=Pt Et ½Ytþ1 : ð4Þ
d d
Denoting gtþ1 as log dividend growth, gtþ1 ¼ logðDtþ1 =Dt Þ, we can rear-
range (4) and iterate forward to obtain the present value relation:
" !#
Pt X 1 X
i1 Xi
d
¼ Et exp tþj þ gtþj , ð5Þ
Dt i¼1 j¼0 j¼1
assuming a transversality condition holds. Note that Equation (5) is

different from the Campbell and Shiller (1988a,b) log linear approxima-
tion for the log price-dividend ratio pt dt ¼ logðPt =Dt Þ:
" #
X
1
j1 d
pt dt c þ Et ðytþj gtþj Þ , ð6Þ
j¼1
where c and are linearization constants. Equation (5) is an exact

expression and involves true expected returns. In contrast, the approx-
imation in Equation (6) involves actual total log returns yt .
Since the price-dividend ratio varies through time, so must some com-
ponent on the RHS of Equation (5). As the discount rate is the sum of the
risk-free rate and a risk premium, time-varying price-dividend ratios or
dividend yields consequently imply that either risk-free rates, risk pre-
miums, or cash flows must be predictable by the dividend yield. Although
we find predictable components in excess returns, the dividend yield
appears to be a strong predictive instrument at short horizons only
when augmented with the short rate. Of course, the nonlinearity in
Equation (5) may make it difficult for linear predictive regressions to
capture these predictable components. In this section, we examine
whether the dividend yield predicts cash flow growth rates or future
interest rates.6
3.1 Dividend growth predictability

Panel A of Table 4 investigates US dividend growth over two samples,
1935–2001 and 1952–2001. Over the longer sample, we find no evidence of
dividend growth predictability. For the shorter sample, high dividend
yields predict future high dividend growth at the one- and four-quarter
6
Goyal and Welch (2003) show that in a Campbell and Shiller (1988a,b) log-linear framework, the
predictive coefficient on the log dividend yield in a regression of the one-period total return on a constant
and the log dividend yield can be decomposed into an autocorrelation coefficient of the dividend yield
and a coefficient reflecting the predictive power of dividend yields for future cash-flows. In Section 4, we
attribute the time variation of the dividend yield into its three possible components—risk-free rates,
excess returns, and cash flows—using a nonlinear present value model.
668
Table 4
Predictability of dividend growth
Univariate
k-qtrs dy4 only r dy4 2 test
Panel A: US S&P data

1935–2001 1 0.0036 0.0586 0.0027 0.974
(0.175) (0.227) (0.147)
4 0.0126 0.0389 0.0133 0.738
(0.504) (0.157) (0.588)
20 0.0100 0.0384 0.0091 0.886
(0.489) (0.299) (0.483)
1952–2001 1 0.0251 0.3541 0.0188 0.000**
(2.476)* (2.191)* (1.599)
4 0.0259 0.1791 0.0228 0.005**
(2.503)* (1.131) (1.903)
20 0.0165 0.0863 0.0132 0.325
(0.927) (0.808) (0.649)
Panel B: pooled across the United States, United Kingdom, and Germany
1953–2001 1 0.1545 0.6991 0.1677 0.000**
(15.54)** (4.296)** (14.16)**
4 0.1552 0.4897 0.1642 0.000**
(14.64)** (3.005)** (13.19)**
20 0.0489 0.2678 0.0555 0.030*
(2.544)* (2.276) (2.452)*
Univariate
k-mths dy12 only r dy12 2 test
Panel C: pooled across United States, United Kingdom, Germany, and France
1975–2001 1 0.0179 0.3248 0.0371 0.616
(0.757) (1.194) (1.286)
12 0.0116 0.3348 0.0082 0.377
(0.440) (1.511) (0.281)
60 0.0077 0.0702 0.0025 0.266
(0.138) (0.359) (0.059)
We estimate regressions of cumulated and annualized k-period ahead dividend growth, on log dividend
yields alone or risk-free rates and log dividend yields together. Panels B and C pool data jointly across
countries, constraining the predictive coefficients to be the same across countries. The 2 test column
reports a p-value for a test that both the risk-free rate and log dividend yield coefficients are jointly equal
to zero. T-statistics in parentheses are computed using Hodrick (1992) standard errors.
*p<0.05.
**p<0.01.
horizons. The magnitude of the coefficients is preserved in the bivariate

regression, but the coefficient is no longer significantly different from zero
at the one-quarter horizon and borderline significant at the one-year
horizon. However, the short rate coefficient is positive and strongly
significant at the one-quarter horizon. The coefficient becomes smaller
and insignificant at longer horizons. The joint tests (across the two
669
coefficients) reject at the 1% level for both the one- and four-quarter
horizons.
Campbell and Shiller (1988a,b) note that the approximate linear relation
(6) implies a link between high dividend yields today and either high future
returns, or low future cash flows, or both. Hence, the positive sign of the
dividend yield coefficient in the short sample is surprising. However, the
Campbell–Shiller intuition is incomplete because it relies on a linear
approximation to the true present value relation (5). Positive dividend
yield coefficients in predictive cash-flow regressions can arise in rational
models. For example, Ang and Liu (2006) show how the nonlinearity of the
present value model can induce a positive dividend yield coefficient.
Menzly, Santos, and Veronesi (2004) show that the dividend yield coeffi-
cient is a function of a variable capturing shocks to aggregate preferences.
Consequently, it changes over time and can take positive values.
In Panels B and C of Table 4, we investigate the relation between
dividend yields and cash-flows for other countries. Panel B pools data
across the United States, United Kingdom, and Germany for the 1953–
2001 sample. Unlike the US post-1952 sample, the dividend yield coeffi-
cients are strongly negative. Because the United Kingdom and German
coefficients are so different from the United States (data not shown), a
pooled result is hard to interpret, and the GMM over-identifying restric-
tions are strongly rejected with a p-value of less than 0.001. Nevertheless,
pooling yields negative, not positive, dividend yield coefficients. The
short rate coefficients are strongly positive and are about twice the
magnitude of the US coefficients (a 1% increase in the short rate approxi-
mately forecasts an annualized 70 basis point increase in expected divi-
dend growth over the next quarter).
Panel C reports coefficients for the MSCI sample. The dividend yield
coefficients are small, mostly negative and never statistically significantly
different from zero. The short rate coefficients are also insignificant,
although they are similar in magnitude to the coefficient found in long-
term US data. There is no general pattern in the individual country
dividend yield coefficients (data not shown): the dividend yield coefficient
in the univariate regression is positive (negative) in the United States and
United Kingdom (France and Germany), with the dividend yield coeffi-
cients retaining the same signs in the bivariate regression in each country.
All in all, we conclude that the evidence for linear cash-flow predictability
by the dividend yield is weak and not robust across countries or sample
periods.7
7
It is conceivable that dividend yields exhibit stronger predictive power for real dividend growth. However,
we find the results for real and nominal growth to be quite similar. In the long US sample, the dividend
yield fails to forecast future ex-post real dividend growth and the coefficients are positive. For the shorter
sample, pooled results across the four countries produce negative coefficients that are actually significant
at short horizons. These results are available upon request. Campbell (2003) also finds analogous results.
670
3.2 Interest Rate Predictability

We examine the possibility that the dividend yield predicts risk-free rates
in Table 5, which reports coefficients of a regression of future annualized
cumulated interest rates on log dividend yields. The persistence of the
risk-free rate causes two econometric problems in linear regressions. First,
because the interest rate and dividend yield are both persistent variables,
the regression is potentially subject to spurious regression bias. To
address this issue, we also report results using the detrended dividend
yield, which is the dividend yield relative to a 12-month moving average
(Campbell 1991, Hodrick 1992). Second, the residuals in the predictive
regression are highly autocorrelated, which means that the use of Hodrick
(1992) standard errors is inappropriate. For a one-period horizon, we use
Cochrane–Orcutt standard errors and generalize the use of this procedure
to panel data in the Appendix. We do not report standard errors for
horizons greater than one period, because the residuals contain both
autocorrelation and moving average effects that cannot be accommo-
dated in a simple procedure.
Table 5 reports that the long sample for the United States shows a
positive effect of the dividend yield on future interest rates. The effect is
economically small (a 1% increase in the log dividend yield predicts an
increase in the one-quarter short rate by 1.7 basis points on an annualized
Table 5
Predictability of risk-free rates
k-qtrs dy4 Detrended dy4
Panel A: US Data
1952–2001 1 0.0171 0.0458
(0.202) (0.416)
4 0.0161 0.0429
20 0.0267 0.0413
Panel B: pooled United States, United Kingdom and Germany

1953–2001 1 0.0170 0.0211
(0.923) (0.765)
4 0.0138 0.0170
20 0.0099 0.0074
k-mths dy12 Detrended dy12
Panel C: pooled United States, United Kingdom, Germany, and France

1975–2001 1 0.0594 0.0407
(1.525) (0.201)
12 0.0588 0.0676
60 0.0619 0.0495
We estimate regressions of cumulated and annualized k-period ahead average risk-free rates by log
dividend yields. The detrended log dividend yield refers to the difference between the log dividend yield
and a moving average of log dividend yields over the past year. Panel c pools data across countries. We
compute Cochrane-Orcutt t-statistics (in parentheses) for a one-quarter horizon.
671
basis next quarter). Using a detrended dividend yield also leads to the
same positive sign. While both effects are statistically insignificant, we
view the relationship between dividend yields and future interest rates as
economically important because interest rates are a crucial component of
a present value relation. From the present value relation (5), we expect a
positive relation between dividend yields and future discount rates. The
interest rate enters the discount rate in two ways. The discount rate is the
sum of the risk-free rate and the risk premium and enters these two
components with opposite signs. It is the first component that gives rise
to the positive relation.
Although not statistically significant, the positive sign of interest rate
predictability by dividend yields is robust. First, omitting the 1990s does
not change the inference, but actually increases the t-statistics. Second, we
also find a positive sign for Germany and the United Kingdom in the
long sample and for all countries in the short sample. In particular, for
MSCI data, the individual coefficients range from 0.036 in the United
States to 0.085 in France at the one-month horizon.8
4. A Present Value Model for Stock Returns

In this section, we present a present value model to shed light on what
kind of discount rate processes are most consistent with the predictability
evidence.
4.1 The model

We start with the basic present value relation in (5) and parameterize the
dynamics of the discount rates and cash flows. We assume that the
continuously compounded risk-free rate rt and log dividend growth gtd
follow the VAR:
Xt ¼
þ Xt1 þ "t , ð7Þ
where Xt ¼ ðrt , gtd Þ and "t IID Nð0,Þ. Let the discount rate, t , in
Equation (4) follow the process:
t ¼ þ Xt þ t1 þ ut , ð8Þ
with ut IID Nð0, 2 Þ and "t and ut are independent. We denote the
individual components of as ¼ ðr , gd Þ .
8
We also examine the predictive power of the dividend yield for ex-post real interest rates, similar to
Campbell (2003). Although the individual coefficients across countries fail to have a consistent sign,
pooled results produce positive coefficients at all horizons, as in the nominal case. However, the
coefficients are not statistically significant.
672
Proposition 4.1. Assuming that Xt ¼ ðrt , gtd Þ follows Equation (7) and
that the log conditional total expected return t follows (8), the price-
dividend ratio Pt =Dt is given by
Pt X 1
¼ expðai þ bi Xt þ ci t Þ, ð9Þ
Dt i¼1
where
1
aiþ1 ¼ ai þ ci þ c2i 2 þ ðe2 þ bi þ ci Þ
2
1

þ ðe2 þ bi þ ci Þ ðe2 þ bi þ ci Þ ð10Þ
2
biþ1 ¼ ðe2 þ bi þ ci Þ

ciþ1 ¼ ci 1,
where e2 ¼ ð0, 1Þ , ai and ci are scalars, and bi is a 2 1 vector. The initial
conditions are given by
1
a1 ¼ e2
þ e2 e2
2
ð11Þ
b1 ¼ e2
c1 ¼ 1:
Proof: See Appendix D.

Proposition 4.1 implies that the dividend yield is a highly nonlinear
function of interest rates, excess returns, and cash flows. Not surprisingly,
if discount rates are persistent, the ci coefficients are negative and higher
discount rates decrease the price-dividend ratio. Analogously, when divi-
dend growth is positively autocorrelated, a positive shock to dividend
growth likely increases the price-dividend ratio, unless it entails an oppo-
site discount rate effect ðgd > 0Þ.
The present value model endogenously generates heteroskedasticity.
While previous studies model returns and dividend yields in finite-order
VAR systems (see, among many others, Hodrick 1992, Campbell and
Shiller 1988a,b, Stambaugh 1999), a VAR cannot fully capture the non-
linear dynamics of dividend yields implied by the present value model. We
can also contrast our present value model with Goetzmann and Jorion
(1993, 1995) and Bollerslev and Hodrick (1996), who either ignore the
cointegrating relation between dividends and price levels that
673
characterizes rational pricing or only develop approximate solutions. In

contrast, we impose cointegration between dividends and prices and our
solution is exact.
Under the special case of constant total returns, that is t ¼ , with
¼ ¼ 2 ¼ 0, and IID dividend growth ðgtd ¼
d þ d "gt Þ, Equation (9)
simplifies to a version of the Gordon model:
Pt expð
d þ 12 2d Þ
¼ :
Dt 1 expð
d þ 12 2d Þ
Another important special case of the model is constant expected excess

returns, where t ¼ þ rt , so ¼ ð1, 0Þ , and ¼ 2 ¼ 0. In this case, the
time variation in total expected returns is all due to the time variation in
interest rates. This is the relevant null model for our excess return
regressions where the expected excess return is constant but the total
expected return varies with the interest rate.
Under the null of constant expected excess returns, the gross total
return Yt ¼ ðPtþ1 þ Dtþ1 Þ=Pt Þ less the gross interest rate expðrt Þ is
Et ½Ytþ1 expðrt Þ ¼ expðrt Þ½expðÞ 1,
so regressing the simple net excess return on the interest rate actually yields a
nonzero coefficient on rt . The scaled expected return, Et ½Ytþ1 = expðrt Þ is
constant and equal to . The predictability regressions typically run in the
literature do not correspond to any of these two concepts, since they use
log returns ~ytþ1 logðYtþ1 Þ rt . In our economy, regressing log returns
onto state variables does not yield zero coefficients because the log excess
return is heteroskedastic. However, we would expect these coefficients to
be small, relative to the null of time-varying expected excess returns (where
t takes the full specification in Equation (8)).
Under the alternative of time-varying discount rates in Equation (8),
total expected returns can depend on both fundamentals (short rates and
dividend growth) and exogenous shocks. The case of ¼ 0 represents
fully exogenous time-varying expected returns. By specifying
t ¼ þ Xt , Equation (8) also nests the case of state-dependent
expected returns.
4.2 Estimation
The estimation of the present value model is complicated by the fact that
in the data, we observe dividends summed up over the past year, but we
specify a quarterly frequency in the model. We estimate the present value
model with simulated method of moments (SMM) (Duffie and Singleton
1993) on US data from January 1952 to December 2001. We provide full
details of the estimation in Appendix E.
674
Panel A of Table 6 reports the VAR parameters. Dividend growth

displays significant positive persistence, and the interest rate has a small
and insignificant effect on dividend growth. The implied unconditional
standard deviation of gtd from the estimation is 0.0173 per quarter. If we
estimate a VAR on ðrt , gtd,4 Þ, the implied unconditional standard devia-
tion of gtd,4 is 0.0156 per quarter. Hence, summing up dividends over the
past four quarters effectively creates a smoother series of dividend
growth compared to the true, but unobservable, cash-flow process.
Panel B presents the parameter estimates for five different discount rate
processes in Equation (8). The first model we estimate (Null 1) is a simple
constant total expected stock return benchmark model. The second model
(Null 2) is our main null model because it imposes constant expected
Table 6
Calibration of the present value model
1=2

rt gtd rt gtd
Panel A: estimates of the VAR for ðrt gtd Þ

rt 0.0010 0.9263 0.0000 0.0026 0.0000
(0.0005) (0.0396) (0.0006)
gtd 0.0053 0.0078 0.5489 0.0036 0.0171
(0.0013) (0.0899) (0.1208) (0.0001) (0.0048)
Null 1 Null 2 Alt 1 Alt 2 Alt 3
Panel B: estimates of the discount rate process

100 2.0168 0.8508 0.0407 0.0476 0.0988
(0.0358) (0.0349) (0.0123) (0.0034) (0.2525)
0.9816 0.9298
(0.0052) (0.0616)
r 1.0000 2.0454 0.0007
(0.4038) (0.0011)
gd 0.0041 0.0014
(0.5704) (0.0024)
100 0.1751 0.5024
(0.0475) (0.1616)
2 test p-value 0.0000 0.0000 0.0101 0.0000 0.1332
The table reports parameter estimates and standard errors in parentheses of the present value model.
Panel A reports estimates of the VAR of short rates and dividend growth in Equation (7). The short rate
rt equation is an AR(1), with standard errors produced by GMM with four Newey–West (1987) lags. The
parameters for gtd are estimated using SMM by matching first and second moments of gtd,4 , along with
the moments Eðrt gtd,4 Þ, Eðgt4
d,4 d,4
gt Þ, and Eðrt4 gtd,4 Þ. Panel B reports parameter estimates for the discount
rate process t ¼ þ Xt þ t1 þ ut [see Equation (8)], with ¼ ðr , gd Þ . The Null Models 1 and 2
impose the restriction ¼ gd ¼ ¼ 0, with r ¼ 0 for Null Model 1 or r ¼ 1 for Null Model 2. These
represent the null hypotheses of constant expected total returns (Null Model 1) or constant expected
excess returns (Null Model 2). Alternative Model 1 sets r ¼ gd ¼ 0, so the discount rate process is
entirely exogenous, whereas Alternative Model 2 imposes ¼ gd ¼ 0, so the discount rate process is
entirely endogenous. In Alternative Model 3, all parameters of the discount rate process are nonzero.
The estimation is done by holding the VAR parameters fixed and matching the first and second moments
of excess returns and dividend yields, along with lagged short rates and lagged dividend growth as
instruments. The last row reports the p-value from a 2 overidentification test.
675
excess returns by setting r ¼ 1 and gd ¼ ¼ ¼ 0. The first model under

the alternative of time-varying discount rates that we consider (Alternative
1) features completely exogenous discount rates ð ¼ 0Þ. The estimation
shows that the log discount rate is very persistent ð ¼ 0:98Þ and its
unconditional variance is about 1% at the quarterly level. Alternative 2
sets ¼ 0 but allows the discount rate process to depend on the two state
variables. We find a slightly positive but insignificant effect of dividend
growth rates on the discount rate, but a strong and significantly negative
interest rate effect. Alternative 3 combines the two models. The negative
interest rate effect disappears, but the zero coefficient means that excess
returns are negatively related to interest rates. The persistence of the
discount rate process now drops to 0.93. The last line of Panel B reports
the p-value of the 2 -test of the overidentifying restrictions for the SMM
estimation. Only Alternative 3 passes this test.
4.3 Economic implications

In this section, we investigate how well the present value models match
the data moments and decompose the variability of the price-dividend
ratio into its components. We also examine how well linear predictive
regressions capture true expected returns implied by the models.
4.3.1 Moments and price-dividend ratio variance decomposition. Table 7

reports a number of implied moments for the various models. Panel A
reports the variance and mean of the dividend yield and excess returns. We
start by examining the null models. Because Null 1 has no time variation in
expected returns, it underestimates the volatility of excess returns and gen-
erates only one-tenth of the dividend yield variability present in the data. The
annualized mean equity premium is only 2.6% instead of 6.12% in the data,
but this is still comfortably within two standard errors of the data moment.
The model also matches the mean dividend yield. The Null 2 model has
similar mean implications, but the variation of excess returns increases
substantially and endogenous dividend yield volatility triples. By definition,
all of the variation in price-dividend ratios should come from either short
rates or cash flows, which we confirm in a variance decomposition of the
price-dividend ratio reported in Panel B.
The variance decompositions represent the computation:
varz ðP=D4 Þ
1 , ð12Þ
varðP=D4 Þ
where varðP=D4 Þ is the variance of price-dividend ratios implied by the

model, and varz ðP=D4 Þ is the variance of the price-dividend ratio pro-
duced by the model when the variable z is nonstochastic and set at its
676
Table 7
Economic implications of the present value model
US data
Null 1 Null 2 Alt 1 Alt 2 Alt 3 Estm SE
Panel A: implied selected moments

Mean excess return 0.0065 0.0066 0.0066 0.0054 0.0068 0.0153 0.0057
Mean dividend yield 0.0335 0.0339 0.0343 0.0348 0.0352 0.0349 0.0008
Volatility excess return 0.0393* 0.0472* 0.0785 0.0824 0.0811 0.0779 0.0061
Volatility dividend yield 0.0010* 0.0031* 0.0120 0.0072* 0.0115 0.0114 0.0011
Panel B: decompositions of the variance of the price-dividend ratio

Percent of short rate 0.0009 0.8745 0.0003 0.9757 0.2221
Percent of dividend growth 1.0000 0.0545 0.0042 0.0328 0.0689
Percent of total discount rate 0.0009 0.1418 0.9955 0.1593 0.9295
Percent of excess return 0.0008 0.0000 0.0214 0.2911 0.6132
Panel C: correlations between true expected excess returns and forecasts from predictive regressions
k ¼ 1, univariate regression dy4 0.7364 0.7783 0.7859
k ¼ 1, bivariate regression dy4 ,r 0.9325 0.9785 0.8654
The table reports various economic and statistical implications from the present value models. Panel A
reports various moments and summary statistics implied from each model. The quarterly moments
reported are the mean and volatility of excess returns and dividend yields in levels. The data standard
errors of the moments are computed by GMM with four Newey–West (1987) lags. In Panel B, the
variance decompositions report the computation 1 varz ðP=D4 Þ=varðP=D4 Þ, where varðP=D4 Þ is the
variance of price-dividend ratios implied by the model, and varz ðP=D4 Þ is the variance of the price-
dividend ratios where all realizations of z ¼ rt , gt , t or the risk premium are set at their unconditional
means. Panel C reports the correlation between fitted values from the excess return predictive regressions
and true conditional expected excess returns Et ð~ytþk Þ implied by the model. The population moments
implied by the model are computed using 1,00,000 simulations from the estimates in Table 6. In Panel A,
the standard errors for the sample moments are computed using GMM.
*Indicates that the population moments lie outside a two standard deviation bound around the point
estimate of the sample moment.
long-run mean. We take z to be short rates, dividend growth, total

discount rates, and excess discount rates, respectively. For Null 2, the
price-dividend ratio variance accounted for by short rate movements is
87.4%. The total discount rate accounts for 14.2% of the variation of the
price-dividend ratio, but this is all due to time-varying interest rates. By
construction, the excess discount rate does not account for any of the
volatility of the price-dividend ratio. Dividend growth also accounts for
only a small part (5.5%) of total price-dividend ratio volatility.
The alternative models 1–3 match the data much better than the null
models, showing that some variation in (excess) discount rates is essential.
In particular, all three alternative models match the variability of excess
returns and, in addition, Alternatives 1 and 3, both featuring exogenous
discount rate variation, match the variability of dividend yields.
677
Alternative 3, the best fitting model, also perfectly matches the first-order
autocorrelation of the dividend yield (0.9596 compared to 0.9548 in US
data spanning 1952–2001). The variance decomposition of the various
models are as expected, given how the discount rates are modeled in each
specification: in Alternative 1, most variation comes from the exogenous
discount rate; in Alternative 2, almost all variation comes from the short
rate (which in turn drives variation in the discount rate), and in Alter-
native 3, the exogenous discount rate dominates but interest rates are still
important. Since Alternative 3 fits the data the best, the variance decom-
position is of considerable interest. It suggests that 61% of the price-
dividend ratio variation is driven by risk premiums, 22% by the short
rate and 7% by dividend growth. The remainder is accounted for by
covariance terms.
4.3.2 How well do predictive regressions capture true expected

returns?
In Panel C of Table 7, we examine how well linear forecasting models
can capture true time variation in expected returns in the alternative
models. We compute the correlation of expected log excess returns,
Et ð~
ytþk Þ, implied by the model with the fitted value on the RHS of the
regression (2).9 There are two striking results in Panel C. First, a uni-
variate dividend regression captures a much smaller proportion of move-
ments in log expected excess returns than a bivariate regression including
both risk-free rates and dividend yields. For example, for Alternative 1
(3), the correlation rises from 74% (75%) in a univariate regression to 93%
(84%) in a bivariate regression at a one-quarter horizon.
Second, long-horizon regressions do not necessarily capture the
dynamics of true expected excess returns better than short-horizon regres-
sions. Cochrane (2001) and others, show that in linear VAR models, long-
horizon regressions more successfully capture predictable components in
expected returns. However, in our nonlinear present value model, long-
horizon regressions may fare worse than short-horizon regressions in
capturing true expected returns. For example, for Alternative 3, the
forecasts from a one-quarter regression with the short rate and dividend
yield have a correlation of 87% with true expected excess returns, whereas
the correlation is 82% at a 20-quarter horizon. In Alternative 2, where
there are no exogenous components in discount rates, the correlation
between the bivariate regression and true expected returns is around
9
In a present value model, the economically relevant quantity for discount rates is the log expected return
t ¼ logððPtþ1 þ Dtþ1 Þ=Pt Þ ¼ logðEt ðYtþ1 ÞÞ. However, the predictive regressions produce a forecast of
expected log returns Et ðlogðYtþ1 ÞÞ. The two quantities logðEt ðYtþ1 ÞÞ and Et ðlogðYtþ1 ÞÞ are
P not equiva-
lent because of time-varying Jensen’s inequality terms. For example, the correlation of tþj rtþj in
Null Model 2 is zero with any variable, whereas the correlation of expected excess log returns Et ð~ytþk Þ is
not. We compute Et ð~ytþk Þ following the method described in Appendix E.
678
98% for all horizons. However, this alternative has the worst fit with the
data.
4.4 Implications for predictive regressions

We now investigate how well the present value models fit the linear
patterns of predictability that we observe in the data. In Table 8, we
compare regression coefficients implied by the present value models to
their values in data.
4.4.1 Expected excess return regressions. In the Null 1 model, total

expected returns are constant. Hence excess returns are, by construction,
Table 8
Predictive regressions implied by the present value model
US data
Null 1 Null 2 Alt 1 Alt 2 Alt 3 Estm SE
Excess return regressions

k ¼ 1, univariate regression dy4 0.0416* 0.0404* 0.0886 0.3564* 0.1864 0.0979 0.0635
k ¼ 1, bivariate regression r 1.0500 0.0824* 0.9310 2.6369 1.0737 2.1623 0.7426
4
dy 0.0049* 0.0214* 0.0878 0.0574 0.2269 0.1362 0.0633
k ¼ 4, univariate regression dy4 0.0269 0.0177 0.0870 0.3122* 0.1723 0.1060 0.0686
k ¼ 4, bivariate regression r 0.9419 0.1191 0.8541 2.5694 0.8562 1.4433 0.7478
4
dy 0.0062 0.0097 0.0862 0.0208 0.2046 0.1313 0.0683
k ¼ 20, univariate regression dy4 0.0213 0.0086 0.0787 0.1772 0.1232 0.0594 0.1245
dy4 0.0040 0.0075 0.0783 0.0004 0.1363 0.0774 0.6445
Dividend growth regressions
k ¼ 1, univariate regression dy4 0.8543* 0.0918* 0.0045* 0.0199 0.0412 0.0251 0.0101
k ¼ 1, bivariate regression r 0.2611 0.1548 0.2942 0.8837* 0.1154 0.3541 0.1616
dy4 0.8452* 0.0561* 0.0047 0.1201* 0.0369 0.0188 0.0118
k ¼ 4, univariate regression dy4 0.9392* 0.0920* 0.0050 0.0259 0.0212 0.0259 0.0103
k ¼ 4, bivariate regression r 0.1478 0.1019 0.1846 0.7031* 0.0997 0.1791 0.1584
dy4 0.9340* 0.1155* 0.0052 0.1057* 0.0174 0.0228 0.0120
k ¼ 20, univariate regression dy4 0.2274* 0.0375 0.0017 0.0016 0.0059 0.0165 0.0178
dy4 0.2247* 0.0408 0.0016 0.0186 0.0032 0.0132 0.1021
Risk-free rate regressions
k ¼ 1, univariate regression dy4 0.0342* 0.2142* 0.0008 0.1052* 0.0354 0.0171 0.0390
k ¼ 4, univariate regression dy4 0.0334 0.1928 0.0008 0.0946 0.0321 0.0161 –
k ¼ 20, univariate regression dy4 0.0230 0.1163 0.0016 0.0568 0.0194 0.0267 –
The table reports implied predictive regression coefficients from the present value models. The LHS
variables are cumulated log excess returns, dividend growth, and risk-free rates. The population moments
are computed using 1,00,000 simulations from the estimates in Table 6. The standard errors for the
regressions in US data are computed using Hodrick (1992) standard errors for the excess return and
dividend growth regressions and using a Cochrane-Orcutt procedure for the k ¼ 1 risk-free rate regres-
sion. All horizons k are in quarters.
*Indicates that the coefficients lie outside a two standard deviation bound around the point estimate of
the sample coefficient.
679
negatively related to interest rates, which explains the large negative short
rate coefficients in the bivariate regressions. Because dividend yields are
correlated with interest rates, the univariate regression also picks up some
predictability. In the Null 2 model, expected excess returns are constant,
so that the only predictability we should observe comes from nonlinea-
rities. As in the Null 1 model, the coefficients on the dividend yield are
invariably small and of the wrong sign compared to the predictions in
Campbell and Shiller (1988a,b).
The alternative models imply different patterns of predictability. Despite
substantial variation in the exogenous discount rate, Alternative 1 generates
a population slope coefficient of only 0.089 in a dividend yield regression,
which is slightly below what is observed in the data. Because the discount
rate is not linked to the interest rate, an excess return projection on
dividend yields and the interest rate leads to coefficients on interest rates
in the neighborhood of 1 and a dividend yield coefficient of 0.088. Both
coefficients are somewhat lower in absolute magnitude than what is
observed in the data.
Alternative 2 implies a dividend yield predictability regression coeffi-
cient of 0.356. This is very close to the dividend yield coefficient for the
United States if we ignore the 1990s (0.296 in Table 2). Hence, in a world
where discount rates only depend on short rates and dividend growth, the
dividend yield would indeed be a strong predictor of excess returns.
However, because the variation in discount rates is mostly driven by
short rates, the dividend yield coefficient drops to 0.057 in a bivariate
regression and the predictable component is now mostly absorbed by the
short rate. This is inconsistent with the data, where bivariate regressions
yield larger dividend yield coefficients, not smaller ones.
Alternative 3 combines features of both Alternatives 1 and 2 and the
coefficients are nicely in between the two alternatives. This model also
yields a negative omitted variable bias in the univariate dividend yield
regression as is true in the data. Note that, with the exception of the
univariate slope coefficient in the dividend yield regression under Alter-
native 2, all predictability coefficients for all alternative models are within
two standard errors of the observed coefficients in the data.
4.4.2 Dividend growth regressions Table 8 also reports how the different
models fare with respect to the predictability of dividend growth. The main
feature in the data is that both dividend yields and short rates receive
positive coefficients in a bivariate regression. Since practically all of the
variation in dividend yields is due to dividend growth in the Null 1 model,
dividend growth is much too predictable in this model. Dividend growth
predictability in the Null 2 model is also inconsistent with the data.
680
In the alternative models, positive dividend yield coefficients in divi-

dend growth regressions occur primarily in two ways. First, the discount
rate can depend positively on dividend growth [gd > 0 in Equation (8)].
Positive persistence in dividend growth normally leads to a negative
association between dividend yields and future cash flows, but in this
case, high dividend growth also increases the discount rate and increases
dividend yields. Second, the short rate may enter negatively in the divi-
dend growth Equation (7). This means that low discount rates (which
tend to raise the dividend yield) are directly associated with relatively
higher expected dividend growth rates. Our estimates for Alternative 3
share both these features (Table 6), and the implied moments of this
model are fully consistent with what is observed in the data.
4.4.3 Risk-free rate regressions A rational present value model with

stochastic interest rates also implies that dividend yields predict interest
rates. Under Null 2, the coefficient is 0.214, much larger than what is
observed in the data. Since interest rates do not affect the risk premium in
this model, we observe a very strong positive relation between the current
dividend yield and future interest rates. The smaller coefficient in the data
indicates that the risk premium component is likely to negatively depend
on interest rates. Hence, the current dividend yield should show a nega-
tive relation with the interest rate component of future risk premiums. In
Alternative 1, the total discount rate does not depend on the interest rate,
yielding a coefficient very close to zero. Alternative 2 makes the interest
rate dependence of the discount rate too strong, resulting in a negative
coefficient. In contrast, Alternative 3 most closely (but not perfectly)
matches the coefficient in the data. Hence, we conclude that the present
value model represented by Alternative 3 is remarkably consistent with
the data.
5. Bias, Size and Power

The inference regarding predictability may critically depend on the finite
sample properties of the estimator. In this section, we investigate the finite
sample bias, size, and power of the estimators in the linear predictive
regressions.
5.1 Small sample bias

In a linear return-dividend yield system (Stambaugh 1999, Amihud and
Hurvich 2004, Lewellen 2004), there is an upward bias in the predictive
coefficient on the dividend yield, deriving from the negative correlation
between return and dividend yield innovations and the persistence of
dividend yields. This analysis does not apply in our framework for two
reasons. First, both the return and the dividend yield are nonlinear
681
Table 9
Small sample bias under null model 2
1975–2001, 1952–2001, 1935–2001, Population,

k-qtrs T ¼ 104 T ¼ 200 T ¼ 267 T ¼1
Scaled returns
Univariate 1 dy4 0.0235 0.0045 0.0012 0.0000
Bivariate 1 r 0.3469 0.1799 0.1435 0.0000
dy4 0.0339 0.0286 0.0264 0.0000
Excess log returns

Univariate 1 dy4 0.0245 0.0065 0.0027 0.0404
4 dy4 0.0408 0.0204 0.0105 0.0177
20 dy4 0.0441 0.0273 0.0185 0.0086
Bivariate 1 r 0.3521 0.1831 0.1388 0.0824
dy4 0.0343 0.0275 0.0295 0.0214
4 r 0.2632 0.1345 0.0989 0.1191
dy4 0.0038 0.0005 0.0057 0.0097
20 r 0.1941 0.1003 0.0696 0.0050
dy4 0.0137 0.0105 0.0067 0.0075
The table reports small sample and population parameter coefficient values for regressions of scaled or
cumulated log excess returns onto log dividend yields (univariate regression) or annualized short rates
and log dividend yields (bivariate regression) from the Null 2 Model. Scaled returns are defined as
Ytþ1 = expðrt Þ, where Ytþ1 is the gross total equity return. The population moments from the model are
computed using 1,00,000 simulations. The small sample moments are computed using 10,000 simulations
of samples of varying length. All horizons k are in quarters.
processes in our model. Second, once we consider multivariate regres-

sions, the bias can no longer be signed in all cases, as already noted by
Stambaugh (1999).
Table 9 reports the small sample bias in several regressions based on
data generated from the Null 2 model (constant expected excess returns).
We consider the small sample distributions for samples of length 104, 200,
and 267 quarters, which correspond to the 1975–2001, 1952–2001, and
1935–2001 sample periods, respectively. We start with regressions using
scaled returns, Ytþ1 = expðrt Þ, as the dependent variable. In these regres-
sions, the population coefficients are zero as the null imposes constant
expected scaled returns. In small samples, the univariate regression yields
a Stambaugh bias that is negligible for our longest sample (0.0012).
Interestingly, in the bivariate regression, the bias on the dividend yield
coefficient is now negative, but the bias on the short rate coefficient is
positive and rather large, ranging from 0.35 for the smallest sample to
0.14 for the longest sample. In this model, the short rate plays the role of
the dividend yield in the linear systems. It is the prime determinant of the
variation in the price-dividend ratio, it is contemporaneously negatively
correlated with returns, and it is very persistent. Hence, a Stambaugh-like
bias for short rates, but not dividend yields, results.
In the second panel of Table 9, we report the bias results for the
predictive regressions using log excess returns. Note that the population
682
coefficients are not zero because of heteroskedasticity. For the univariate

dividend yield regressions, the population coefficients are slightly nega-
tive at short horizons and positive at the five-year horizon. For our small
sample, they interact with the Stambaugh bias to produce a small upward
bias for our smallest sample but a downward, negligible bias for our
longest sample. For the bivariate regressions, the population biases are
negative for both the short rate and dividend yield coefficients. However,
in small samples, the scaled return biases clearly dominate and we find the
short rate (dividend yield) coefficient to be biased upward (downward).
The small sample biases become significantly smaller at long horizons and
turn slightly positive for the dividend yield coefficient. Clearly, recent
inference techniques that focus on univariate regressions in a linear
framework [Valkanov (2003), Torous, Valkanov, and Yan (2004), Polk,
Thompson, and Vuolteenaho (2006)] are of little use in our setting.
In summary, the small sample biases under the null strengthen our
empirical evidence. That is, the univariate dividend yield predictability
coefficients are slightly overestimated, and the small sample biases for the
bivariate coefficients go the wrong way at short horizons: the estimated
short rate coefficient is negative, but the small sample bias is positive; and
the estimated dividend yield coefficient is positive, but the small sample
bias is negative.
5.2 Size
Table 10 reports empirical sizes for tests of a 5% nominal (asymptotic)
size. In the shortest sample for the one-quarter horizon, the univariate
dividend yield regression displays negligible size distortions, but for the
bivariate regressions, all tests slightly over-reject at asymptotic critical
values. For longer horizons, the performance of the Newey–West and
robust Hansen–Hodrick estimators deteriorates, with the empirical size
exceeding 33% for a 5% test in the univariate dividend regression.10 For
our longer samples, these distortions become smaller but do not disap-
pear. For example, at the five-year horizon, the empirical size of the
dividend yield regression still exceeds 18.5% for both Newey–West and
robust Hansen–Hodrick estimators in the 267 quarter sample.
The Newey–West and robust Hansen–Hodrick standard errors are too
small because they underestimate the serial correlation in the error terms
as the autocorrelation estimates are downwardly biased. The Newey–West
standard error also uses a Bartlett kernel of declining, tent-shaped weights.
Under the null, the kernel is rectangular, so the Newey–West standard
errors underweight the effect of autocorrelations at long lags unless a higher
10
Hodrick standard errors also have relatively few size distortions, especially compared to Newey-West
standard errors, for standard linear VARs, like the Stambaugh (1999) system. Hodrick (1992) also
demonstrates that Hodrick standard errors are correctly sized for multivariate VARs.
683
684

Table 10
Empirical size
Univariate regression Bivariate regression
2 tests
dy4 r dy4 joint across horizon
Robust Robust Robust

k-qtrs N–W H–H Hodrick N–W H–H Hodrick N–W H–H Hodrick Univariate Bivariate
Sample length=104 quarters (1975–2001)

1 0.053 0.047 0.047 0.076 0.070 0.070 0.065 0.059 0.059 0.051 0.010
4 0.132 0.104 0.044 0.151 0.138 0.062 0.113 0.109 0.050
20 0.338 0.373 0.047 0.335 0.349 0.060 0.232 0.244 0.037

1 0.045 0.042 0.042 0.057 0.054 0.054 0.059 0.054 0.054 0.056 0.012
4 0.112 0.077 0.044 0.109 0.092 0.051 0.095 0.084 0.046
20 0.226 0.230 0.043 0.205 0.231 0.051 0.129 0.156 0.039

1 0.044 0.041 0.041 0.062 0.058 0.058 0.057 0.055 0.055 0.052 0.013
4 0.100 0.069 0.040 0.096 0.079 0.046 0.087 0.075 0.044
20 0.188 0.186 0.039 0.173 0.194 0.048 0.110 0.138 0.038
The table lists empirical size properties corresponding to a nominal size of 5% of Newey–West (1987) (N–W) with k þ 1 lags, Robust Hansen–Hodrick (1980) (Robust H–H) and
Hodrick (1992) t-statistics. We examine a univariate regression of excess returns on dy4 and a bivariate regression of excess returns on r and dy4 . We simulate 10,000 samples of
various lengths from the Null 2 Model (constant expected excess returns) and record the percentage of observations greater than the nominal critical values under the null
hypothesis of no predictability. The 2 tests report the proportion of rejections (using a 5% nominal size) for testing predictability jointly across horizons using Hodrick
standard errors. In the case of the bivariate regression, joint predictability of both short rates and log dividend yields is considered across horizons. All horizons k are in
quarters.
lag length is chosen. While the robust Hansen–Hodrick standard errors

employ the correct rectangular filter, the estimate of the covariance matrix
is not guaranteed to be invertible in small samples. In contrast, Hodrick
uses covariance stationarity to remove the MA structure in the residuals.
Hodrick’s covariance estimator in Equation (3) is invertible and avoids the
biased estimation of autocorrelations at long lags.
For Hodrick standard errors, the worst size distortion occurs at the
one-quarter horizon with a 7% empirical size for a 5% nominal test on the
short rate coefficient. For longer horizons, the Hodrick tests become
slightly conservative, with the worst size distortion occurring for the
dividend yield coefficient in the bivariate regression (a 3.7% empirical
size for a 5% nominal test). The joint tests across horizons show few size
distortions for the univariate regression, but are somewhat too conserva-
tive for the bivariate regression. In summary, the Hodrick standard errors
display very satisfactory small sample properties that are far superior to
those of Newey–West and Hansen–Hodrick standard errors.
5.3 Power
5.3.1 Power in one country. Table 11 summarizes the size-adjusted
power for the Hodrick standard errors under Alternatives 1–3. Alterna-
tive 1 generates about the same predictive coefficient on the dividend yield
as observed in the data but the predictive power of the short rate is
unrealistically weak. Panel A shows that the power of the univariate
dividend yield regression is very small for the shortest sample, only
25.5% (12.1%) at the one-quarter (20-quarter) horizon. For the longer
samples, power rises to around 60% (40%) at the one-quarter (20-quarter)
horizon. The results for the bivariate regressions are quite similar, with
the power for the short rate coefficient slightly weaker than for the
dividend yield. Hence, power is only satisfactory for the long samples,
and even here, we might fail to reject the null of no predictability even
though predictability is truly present.
Alternative 2, reported in Panel B, generates as much short rate pre-
dictability as in the data but slightly overpredicts the univariate predictive
power of the dividend yield. For the 104-quarter length sample at short
horizons, the power of the univariate dividend yield regression is satisfac-
tory (55.2%). The power deteriorates with the horizon reaching 12.6% at
five years. Power increases substantially with the sample size; for our
longest sample, the Hodrick test has a power of over 97% at the one-
quarter horizon. If predictability in the data is as strong as under Alter-
native 2, it is unlikely that we failed to detect univariate predictability.
The power to detect the predictive ability of the short rate in the bivariate
regression is very high for all samples and at all horizons. In the bivariate
regression, there is little power to detect the true relation of future excess
returns with the dividend yield, because the dividend yield coefficient is
685
Table 11
Size-adjusted power for Hodrick standard errors for one country
Univariate Bivariate 2 tests

regression regression joint across horizon
Sample Horizon
length (Qtrs) k-qtrs, dy4 r dy4 Univariate Bivariate
Panel A: power under alternative 1: t ¼

þ t1 þ ut
104 1 0.255 0.201 0.283 0.325 0.335
4 0.232 0.189 0.268
20 0.121 0.096 0.156
200 1 0.625 0.415 0.650 0.520 0.557
4 0.556 0.410 0.608
20 0.395 0.248 0.436
267 1 0.596 0.344 0.596 0.632 0.691
4 0.548 0.338 0.589
20 0.391 0.161 0.431
Panel B: power under alternative 2: t ¼
þ Xt
104 1 0.552 0.636 0.075 0.559 0.036
4 0.463 0.795 0.041
20 0.126 0.837 0.007
200 1 0.892 0.795 0.091 0.891 0.042
4 0.808 0.942 0.048
20 0.362 0.984 0.004
267 1 0.971 0.853 0.094 0.967 0.048
4 0.940 0.975 0.054
20 0.577 0.998 0.004
Panel C: power under alternative 3: t ¼
þ t1 þ Xt þ ut
104 1 0.566 0.138 0.628 0.587 0.617
4 0.484 0.090 0.562
20 0.159 0.016 0.198
200 1 0.874 0.243 0.902 0.866 0.905
4 0.775 0.181 0.859
20 0.417 0.021 0.483
267 1 0.959 0.292 0.971 0.952 0.976
4 0.923 0.225 0.951
20 0.620 0.024 0.682
The table lists empirical power properties corresponding to a size-adjusted level of 5% of Hodrick (1992)
t-statistics. We examine a univariate regression of excess returns on dy4 and a bivariate regression of
excess returns on r and dy4 . We simulate 10,000 samples of various lengths from Alternatives 1–3 (Panels
A–C) and record the percentage of observations greater than the 5% critical values recorded under the
Null 2 Model (constant expected excess returns) using the simulations in Table 10. The 2 tests report the
proportion of rejections (using a 5% nominal size) for testing predictability jointly across horizons of
both short rates and log dividend yields. All horizons k are quarterly.
very small in the presence of the short rate as a predictor. Consistent with
this, the univariate joint 2 tests across horizons is very powerful, but the
power of the test in the bivariate regression is minimal.
Panel C reports the power under Alternative 3, which fits the data the
best. The power of the univariate tests is slightly better than the power
under Alternative 2. For the bivariate regression, the predictive coeffi-
cient of the short rate (dividend yield) is underestimated (overestimated)
686
relative to the data coefficients under Alternative 3. Consequently, the

power to detect the true predictive relation with the short rate is low
at long horizons (because the predictability through the short rate is
short-lived), whatever the sample size is. Power at a one-quarter
horizon rises from 13.8% for the shortest sample to 29.2% for the
longest sample. Fortunately, the power to detect predictability
through the dividend yield in the bivariate system is quite high:
power is 62.8% (97.1%) for the shortest (longest) sample at a one-
quarter horizon. For a 20-quarter horizon, we need longer samples to
obtain a powerful test (power is only 19.8% for the 1975–2001 sample
but is 68.2% for the 1935–2001 sample). With such high power, it is
unlikely that we would have failed to uncover predictability by the
dividend yield. Overall, the joint tests are quite powerful (at least
58.7% power). We conclude that lack of power is unlikely to drive
our failure to find predictability of excess returns by dividend yields,
particularly for the longer US sample, unless the true predictability in
the data is quite weak to begin with.
5.3.2 Power pooling cross-country data To examine the increase in

power that pooling cross-country data allows, our data generating
process must match the empirical correlations of excess returns across
countries in the data. For the United States, United Kingdom, France,
and Germany, the cross-country correlations for excess returns are all
around 0.50 (Table 1). To account for the cross-sectional correlation, we
modify the present value model in the following fashion. First, we
consider each country to be a separate draw of ðXt , t Þ, using Alterna-
tive 3. Second, we specify the process for Xt in each country to be
independent but allow the discount rates for different countries to be
correlated. Specifically, we allow the shocks u it in the discount rate (8)
for country i to be correlated with the discount rate shocks u jt for
country j. We set this correlation at 0.80, which produces an uncondi-
tional correlation coefficient of 0.535 between the implied excess equity
returns for any two countries.
To compare the increase in power pooling international data, Table 12
reports power under Alternative 3 for a cross-sectional panel of N coun-
tries and compares it to increasing the short US sample by N times.
Pooling data by adding one other country turns out to be only slightly
worse than doubling the sample size of the United States. For four
countries, the pooled test is better than using a sample of 312 quarters,
which is larger than the longest US sample we have. Hence, pooling
information across countries should increase the confidence in our
results more than the US long sample does from a pure statistical
power perspective.
687
Table 12
Power properties for pooling cross-country data
One country Multiple countries
Univariate Univariate
regression Bivariate regression regression Bivariate regression
4 4 2 4
k-qtrs dy r dy test dy r dy4 2 test
Sample length=104 qtrs

1 0.575 0.096 0.612 0.487
4 0.509 0.058 0.536 0.412
20 0.201 0.018 0.169 0.133
Sample length = 208 qtrs Two countries
1 0.891 0.193 0.913 0.839 0.887 0.143 0.905 0.835
4 0.847 0.115 0.878 0.779 0.827 0.070 0.847 0.747
20 0.503 0.008 0.499 0.350 0.326 0.009 0.327 0.227
Sample length = 312 qtrs Three countries
1 0.980 0.289 0.986 0.967 0.974 0.218 0.978 0.963
4 0.962 0.179 0.977 0.941 0.952 0.111 0.963 0.922
20 0.745 0.010 0.762 0.603 0.556 0.009 0.590 0.446
Sample length = 416 qtrs Four countries
1 0.998 0.387 0.999 0.995 0.995 0.293 0.996 0.991
4 0.993 0.255 0.996 0.989 0.988 0.153 0.991 0.978
20 0.881 0.012 0.898 0.790 0.739 0.013 0.767 0.652
The table lists empirical power properties of Hodrick (1992) t-statistics, comparing a sample of one country of
increasing length versus a cross-sectional panel of countries, each of length 104 quarters. (The number of
observations 104 quarters corresponds to the sample period 1975–2001.) Power is taken corresponding to a
nominal (asymptotic) size of 5% using Alternative 3. We examine a univariate regression of excess returns on
log dividend yields and a bivariate regression of excess returns on short rates and log dividend yields. The 2
test reports a test for the joint predictability of the short rate and log dividend yield for a given horizon. The
population correlation coefficient between the excess returns of any two countries is 0.535.
6. The Predictive Power of the Earnings Yield

Our findings so far suggest that dividend yields have marginal predictive
power for returns within a bivariate regression, with the short rate dom-
inating the dividend yield. Campbell and Shiller (1988a) and Lamont
(1998) claim that the earnings yield has information over and above the
dividend yield in capturing predictable components in returns.
6.1 The earnings yield and excess return predictability

The first column of Panel A, in Table 13 reports a univariate regression
with the earnings yield as the regressor. The results are similar to what we
found for the dividend yield regression in Table 2. This suggests that the
weak univariate relation between returns and yield variables primarily
comes from the price in the denominator of both variables.
To allow comparison with Lamont (1998), we report a bivariate regres-
sion of excess returns on log dividend and log earnings yields. Lamont
688
Table 13
Predictability of excess returns by earnings yields
Univariate regression Lamont regression Trivariate regression 2 tests
k-qtrs ey4 dy4 ey4 r dy4 ey4 Lamont Trivariate
Panel A: US quarterly S&P Data

1935–2001 1 0.0665 0.1842 0.1019 1.0334 0.1000 0.0168 0.183 0.119
(1.258) (1.330) (0.741) (1.296) (0.605) (0.103)
4 0.1018 0.1119 0.0012 0.7759 0.0472 0.0654 0.108 0.106
(1.961)* (0.981) (0.011) (1.084) (0.377) (0.556)
20 0.0719 0.1111 0.0092 0.4190 0.0666 0.0291 0.347 0.260
(1.415) (0.946) (0.128) (0.682) (0.486) (0.361)
1952–2001 1 0.0646 0.1948 0.1113 2.6243 0.0337 0.1272 0.192 0.006**
(0.993) (1.557) (0.873) (2.854)** (0.239) (0.815)
4 0.0798 0.1615 0.0636 1.8146 0.0479 0.1029 0.277 0.056
(1.218) (1.195) (0.450) (2.063)* (0.687) (0.687)
20 0.0223 0.1458 0.0765 0.3548 0.1199 0.0418 0.854 0.872
(0.320) (0.561) (0.520) (0.489) (0.432) (0.249)
1935–1990 1 0.0896 0.3848 0.1800 0.6544 0.3051 0.1125 0.049* 0.050*
(1.304) (1.849) (1.062) (0.696) (1.112) (0.499)
4 0.1473 0.2869 0.0532 0.4708 0.2300 0.0046 0.007** 0.018*
(2.309)* (1.872) (0.411) (0.031) (1.336) (0.031)
20 0.1178 0.1916 0.0139 0.4239 0.1423 0.0297 0.017* 0.041*
(2.392)* (1.874) (0.181) (0.714) (1.597) (0.431)
1952–1990 1 0.1028 0.9349 0.5240 2.0892 0.6798 0.2418 0.000** 0.000**
(1.168) (3.710)** (2.618)** (2.096)* (2.338)* (0.949)
4 0.1287 0.8201 0.4212 1.3320 0.6565 0.2392 0.000** 0.001**
(1.521) (3.234)** (2.080)* (1.376) (2.303)* (0.960)
20 0.0798 0.4167 0.1999 0.3068 0.3793 0.1570 0.010** 0.028*
(1.133) (2.685)** (1.570) (0.374) (2.188)* (0.991)
689
690

Table 13
(continued)
k-qtrs ey4 dy4 ey4 r dy4 ey4 Lamont Trivariate
k-mths ey12 dy12 ey12 r dy12 ey12 Lamont Trivariate
Panel B: pooled-country monthly MSCI data

1975–2001 1 0.0251 0.0730 0.0192 1.82d23 0.1609 0.0035 0.612 0.000**
(0.498) (0.969) (0.366) (2.792)** (2.001)* (0.070)
12 0.0098 0.0994 0.0680 1.0401 0.1493 0.0548 0.385 0.000**
(0.195) (1.178) (1.280) (1.956) (1.697) (1.116)
60 0.0369 0.0877 0.0710 0.2956 0.0686 0.0739 0.132 0.321
(0.746) (0.882) (1.980)* (0.731) (0.774) (2.370)*
We estimate regressions of the form ~ytþk ¼ þ zt þ tþk,k , where ~ytþk is the cumulated and annualized k-period ahead return, with the instruments zt being log earnings yields
(univariate regression), or log dividend yields and log earnings yields (Lamont bivariate regression), or risk-free rates, log dividend yields, and log earnings yields (trivariate
regression). T-statistics in parentheses are computed using Hodrick (1992) standard errors. For Panel A (B), horizons k are quarterly (monthly). Panel B pools coefficients
jointly across the United States, United Kingdom, France, and Germany, constraining the coefficients to be the same across countries. The 2 test columns report a p-value for
a test that all the coefficients in each regression are jointly equal to zero.
*p<0.05.
*p<0.01.
finds a positive coefficient on the dividend yield and a negative coefficient

on the earnings yield. He argues that the predictive power of the dividend
yield stems from the role of dividends in capturing permanent compo-
nents of prices, whereas the negative coefficient on the earnings yield is
due to earnings being a good measure of business conditions. Table 13
summarizes that the coefficients over the long US sample, while having
the same sign found by Lamont, are insignificant and not one joint test of
the predictive power of dividend and earnings yields is significant at the
10% level. Only when the 1990s are excluded, as in the 1952–1990 sample
similar to Lamont’s paper, do we find significant coefficients for dividend
and earnings yields. When we add the short rate as a predictor in a
trivariate regression of excess returns on risk-free rates, dividend and
earnings yields, the coefficients on dividend and earnings yields remain
insignificantly different from zero, and the sign on the earnings yield is
fragile. For the post-1952 samples, the short rate predictive power
remains robust in the presence of the earnings yield.
Panel B of Table 13 pools all the specifications across the United States,
United Kingdom, France, and Germany. There are no significant coeffi-
cients in the univariate regression. The bivariate specification preserves
the Lamont coefficient pattern with the earnings yield coefficient now
significantly negative at the five-year horizon. However, joint 2 predict-
ability tests fail to reject the null of no predictability. In the trivariate
specification, the short rate and dividend yield coefficients have similar
signs and significance to the results from the bivariate regressions in
Section 2. The earnings yield coefficient is significantly negative only at
the five-year horizon, but a joint test across all predictors at the five-year
horizon fails to reject the null of no predictability. For individual coun-
tries, the Lamont coefficient pattern is neither robust nor significant
(data not shown). In conclusion, there is little evidence that earnings
yields predict excess returns. The earnings yields coefficients are not
robust across different sample periods or countries.
6.2 The earnings yield and cash-flow predictability

In this section, we use both dividend and earnings yields to predict cash flows,
as measured by dividend growth and earnings growth. Panel A of Table 14
reports the results for dividend growth. In the univariate regression, the
earnings yield is a better predictor of dividend growth than the dividend
yield. The coefficients are larger in magnitude than the dividend yield coeffi-
cients, and, except for the coefficient in the 20-quarter regression, significantly
different from zero. The earnings yield coefficients are all positive.
The bivariate dividend yield–earnings yield regression shows an intri-
guing result. The Lamont (1998) pattern of a positive dividend yield
coefficient and a negative earnings yield coefficient is reversed, and highly
significant, for the cash-flow regressions! At short horizons, both
691
692

Table 14
Predictability of cash-flow growth by earnings yields
Univariate
regression Lamont regression Trivariate regression 2 tests
k ey4 dy4 ey4 r dy4 ey4 Lamont Trivariate
Panel A: predictability of dividend growth

US quarterly S&P data
1935–2001 1 0.0594 0.1897 0.2328 0.9766 0.2694 0.3132 0.002** 0.001**
(2.936)** (2.724)** (3.175)** (3.527)** (3.412)** (3.785)**
4 0.0370 0.1494 0.1713 0.8333 0.2188 0.2402 0.031* 0.013**
(2.061)* (2.021)* (2.377)* (2.877)** (2.559)* (2.874)**
20 0.0062 0.0456 0.0395 0.1428 0.0608 0.0526 0.106 0.194
(0.493) (1.260) (1.985)* (0.835) (1.310) (1.821)
Pooled-country monthly MSCI data
1975–2001 1 0.0214 0.0792 0.0695 0.2059 0.0891 0.0669 0.002** 0.004**
(1.183) (2.741)** (3.396)** (0.775) (2.783)** (3.365)**
12 0.0047 0.0332 0.0241 0.3872 0.0146 0.0290 0.447 0.161
(0.254) (1.063) (1.216) (1.769) (0.440) (1.527)
60 0.0132 0.0090 0.0167 0.0447 0.0118 0.0163 0.413 0.580
(0.570) (0.277) (1.319) (0.262) (0.308) (1.355)
Panel B: predictability of earnings growth

US quarterly S&P data
1935–2001 1 0.0582 0.0970 0.1468 0.0355 0.0941 0.1439 0.391 0.550
(1.120) (0.681) (1.079) (0.049) (0.582) (0.895)
4 0.0822 0.1994 0.2613 0.1563 0.1864 0.2484 0.026* 0.057
(1.791) (1.467) (2.220)* (0.223) (1.235) (1.824)
20 0.0674 0.0876 0.1313 0.1797 0.1066 0.1478 0.007** 0.004**
(1.457) (0.659) (1.901) (0.357) (0.671) (1.695)
Pooled-country monthly MSCI data
1975–2001 1 0.2767 0.2723 0.4419 0.2867 0.2861 0.4383 0.000** 0.000**
(8.381)** (3.566)** (5.528)** (0.743) (3.800)** (5.109)**
12 0.3373 0.3452 0.5392 1.9369 0.4381 0.5148 0.000** 0.000**
(9.787)** (4.334)** (7.153)** (5.473)** (5.627)** (6.883)**
60 0.1534 0.2332 0.2440 0.0974 0.2269 0.2449 0.000** 0.000**
(3.321)** (2.356)* (5.357)** (0.3403) (2.565)* (6.053)**
We estimate regressions of the form ~ytþk ¼ þ z¢t þ tþk,k , where y~tþk is the cumulated and annualized k-period dividend or earnings growth, with the instruments zt being log
earnings yields (univariate regression), or log dividend yields and log earnings yields (Lamont bivariate regression), or risk-free rates, log dividend yields, and log earnings yields
(trivariate regression). T-statistics in parentheses are computed using Hodrick (1992) standard errors. For the quarterly US S&P data, horizons k are quarterly, whereas for the
monthly MSCI data horizons k are monthly. The pooled-country panels pool coefficients jointly across the United States, United Kingdom, France, and Germany,
constraining the coefficients to be the same across countries. The 2 test columns report a p-value for a test that all the coefficients in each regression are jointly equal to zero.
*p<0.05.
**p<0.01.
693
coefficients are different from zero at the 1% level and the joint test also
rejects at the 1% level. The coefficients become smaller, and the statistical
significance weakens, with longer horizons. We still find predictability at
the 5% level for the one-year horizon, but only the earnings yield is
significant (at the 5% level) for the five-year horizon. Adding the risk-
free rate in the trivariate regression does not change this picture very
much, but a high short rate is also a very strong signal of lower future
dividend growth at the one- and four-quarter horizons. One possible
interpretation is that since interest rates are high during recessions and
recessions are persistent (Ang and Bekaert, 2002), high interest rates
predict low future cash flows. The negative dividend yield coefficient
can arise from prices reflecting future positive cash-flow prospects.
We find that the negative dividend yield and positive earnings yield
coefficients for predicting dividend growth are very robust across differ-
ent subsamples, but we do not report these results to conserve space. In
particular, omitting the 1990s, the signs of the coefficients are the same,
but the t-statistics are even larger in magnitude than the those reported
for the full sample. The inverse Lamont pattern is also robust in the
subsamples beginning in 1952.
When we pool data across countries using MSCI data, the cross-
sectional variation in the coefficients makes the cash-flow Lamont pattern
weaker, but it remains significant at the 1% level at the one-month
horizon, in both the bivariate and trivariate regressions. At longer hor-
izons, earnings and dividend yields do not predict future cash flows.
Moreover, the pattern is also repeated in the coefficients for each indivi-
dual country. Internationally, the short rate is not a robust predictor of
future cash flows, perhaps because the cyclicality of interest rates is not
consistent across countries.
In Panel B of Table 14, we repeat the same regressions for earnings
growth. The univariate regression with the earnings yield delivers a nega-
tive coefficient. The sign of this point estimate could be potentially con-
sistent with a standard price effect. However, it is statistically insignificant.
Interestingly, the sign of the dividend and earnings yield coefficients are
reversed for earnings growth, compared to dividend growth: for earnings
growth, we see a positive (negative) coefficient on the dividend (earnings)
yield. The effect is strongest for the earnings yield, which is significant at
the 5% level for the one- and five-year horizons, although borderline in the
latter case. Looking at the joint 2 tests, we conclude that there is some
evidence of cash-flow growth predictability at longer horizons (the p-value
is at most 0.057), primarily driven by the earnings yield. The 1952–2001
sample also preserves these coefficient patterns.
When we pool data across countries, the coefficients increase in mag-
nitude and the statistical significance increases considerably. All the joint
tests now reject at the 1% level, and even the short rate predicts cash-flow
694
growth significantly at the one-year horizon. The strong international

results are due to the fact that in each individual country, the dividend
yield (earnings yield) coefficients are positive (negative).
We conclude that high dividend yields signal high future earnings
growth and high earnings yields signal low future earnings growth rates.
It is conceivable that the second effect is a genuine price effect (higher
prices in response to predicted rises in future earnings), but the first effect
is harder to interpret. The puzzling nature of our findings becomes more
apparent when we look at the earnings and dividend growth simulta-
neously. We find the following sign pattern:
Dividend growth Earnings growth Payout ratio
Dividend yield ve +ve ve
Earnings yield +ve ve +ve
We can derive the results in the last column because the change in the
payout ratio equals (logarithmic) dividend growth minus earnings
growth. Hence, our results imply that higher dividend yields (higher
earnings yields) strongly predict lower (higher) payout ratios tomorrow.
Can we explain the reverse patterns for the two cash-flow measures for
dividend and earnings yields used jointly in predictive regressions? We
can rule out a Lamont (1998) story translated to cash flows. According to
Lamont, dividend yields capture price effects, whereas earnings yields
capture the cyclical component in earnings and hence potentially also
risk aversion. Under this scenario, we would expect the dividend yield to
be negatively related to earnings growth (the usual positive cash flow
prospects), but we find a positive effect.
When dividend yields are high today, we predict low dividend growth in
the future because payout ratios strongly decrease. This may be the result of
dividend smoothing, or it may reflect prices anticipating higher growth
opportunities that decrease the payout ratio. The positive relation between
current high dividend yields and future earnings growth implies that these
growth opportunities do not rapidly translate into higher future earnings.
The negative relation between the current earnings yield and future earnings
may be consistent with either a price effect or mean reversion in earnings.
The payout ratio reacts positively to an increase in the earnings yield. In the
mean reversion story, this could be an artifact of dividend smoothing. In the
price story, lower prices today may reflect poor future earnings and poor
future growth opportunities. The poor growth opportunities may increase
the payout ratio, particularly if dividends are sticky.
All in all, we find very strong evidence of dividend growth predictabil-
ity, both in US and international data, by using the dividend and earnings
yield jointly as predictors. We find weaker evidence in US data, but very
strong international evidence, of earnings growth predictability by
695
dividend and earnings yields. A challenge for future work is to create a

present value model with sophisticated dynamics for earnings growth,
payout ratios, and dividend growth to match this evidence.
7. Conclusion
The predictable components in equity returns uncovered in empirical
work over the last 30 years have had a dramatic effect on finance
research. Theoretical equilibrium models try to match the predictability
evidence as a stylized fact. The partial equilibrium dynamic asset alloca-
tion literature investigates the impact of the predictability on hedging
demands. Much of the focus has been on the predictive prowess of the
dividend yield, especially at long horizons. In this article, we pose the
question whether this predictability exists. After carefully accounting for
small sample properties of standard tests, our answer is surprising but
important. At long horizons, excess return predictability by the dividend
yield is not statistically significant, not robust across countries, and not
robust across different sample periods. In this sense, the predictability
that has been the focus of most recent finance research is simply not there.
Nevertheless, we do find that stock returns are predictable, calling for a
refocus of the predictability debate in four directions. First, our results
suggest that predictability is mainly a short-horizon, not a long-horizon,
phenomenon. The predictive ability of the dividend yield is best seen in a
bivariate regression with short rates only at short horizons. Second, the
strongest predictability comes from the short rate rather than from the
dividend yield. The result that the short rate predicts equity returns goes
back to at least Fama and Schwert (1977), but somehow recent research
has failed to address what might account for this predictability and has
mostly focused on the dividend yield as an instrument. Third, high
dividend yields predict high future interest rates. Finally, dividend and
earnings yields have good predictive power for future cash-flow growth
rates but not future excess returns. Hence, a potentially important source
of variation in price-earnings and price-dividend ratios is the predictable
component in cash flows. Our results generally imply that univariate
linear models of expected returns are unlikely to satisfactorily capture
all the predictable components in returns.
After our results were first distributed in a working paper article (Ang
and Bekaert, 2001), a number of articles have been written that confirm
them. Campbell and Yogo (2006) develop a new inference methodology
within the linear regression framework of Stambaugh (1999) and find that
the predictive power of the dividend yield is considerably weakened but
that the predictive power of the short rate is robust. Lettau and Ludvigson
(2005) also find that the price-dividend ratio weakly forecasts excess returns
but confirm that future dividend growth, long ignored by the literature, is
696
predictable. Engstrom (2003) confirms our findings that univariate dividend

yield regressions have difficulty capturing all the predictable components in
returns, by constructing economies where the dividend yield is a noisy
predictor of both excess returns and cash-flow growth.
We hope that our results will, in the short run, affect the asset alloca-
tion literature, which often has taken the predictive power of the dividend
yield in a univariate regression as a stylized fact, and in the longer run will
stimulate research on theoretical models that might explain the predict-
ability patterns we demonstrate, particularly return predictability by the
short rate at short horizons and the joint predictability of cash flows and
excess returns. Finally, future research should also reconcile the weak
out-of-sample evidence of predictability (Bossaerts and Hillion 1999,
Goyal and Welch 2003, 2004) with the in-sample evidence of predictabil-
ity. While such a reconciliation may require models with structural breaks
or regime shifts (Timmermann and Paye 2006), we note one last interest-
ing result derived from our US sample. Mimicking Goyal and Welch’s
(2004) procedure for out-of-sample forecasting over the post-1964 period,
we find that the bivariate predictive regression with the short rate and
dividend yield produces a lower root mean squared error than the historical
mean for forecasting excess returns at a one-quarter horizon but not at long
horizons. Consequently, for the bivariate predictive regression, the in-
sample and out-of-sample evidence of return predictability is consistent.
Appendix A: Robust Hansen–Hodrick (1980) Standard Errors

Using
p ffiffiffiffi GMM, the parameters ¼ ð k Þ in Equation (2) have an asymptotic distribution
T ð^ ÞNð0,Þ
a
where ¼ Z01 S0 Z01 , Z0 ¼ Eðxt xt Þ, and xt ¼ ð1 z t Þ. We estimate S0
by
X
k1
S^0 ¼ Cð0Þ þ ½CðjÞ þ CðjÞ , ðA1Þ
j¼1
where
1 X T
CðjÞ ¼ ðwtþk wtþkj Þ
T t¼jþ1
and wtþk ¼ "tþk,k xt . This estimator of S0 is not guaranteed to be positive semi-definite. If it

is not, we use a Newey–West (1987) estimate of S0 with k lags. Note that for k ¼ 1, the
robust Hansen–Hodrick (1980) and Hodrick (1992) standard errors are identical.
Appendix B: Testing Predictability Across Horizons

To test whether the predictability coefficients are statistically significant across n horizons
k1 … kn , we set up the simultaneous equations:
697
y~tþk1 ¼ k1 þ k1 zt þ utþk1

... ðB1Þ
y~tþkn ¼ kn þ kn zt þ utþkn :
We produce an estimate ^ of ¼ ðk1 k1 …kn kn Þ by performing OLS on each equation.

The moment conditions for the system in Equation (B1) are
0 1 0 1
htþk1 utþk1 xt
B .. C ¼ EB .. C
EðhtþkÞ E@ . A @ . A ¼ Eðutþk xt Þ ¼ 0, ðB2Þ
htþkn utþkn xt
where xt ¼ ð1 z t Þ, a K 1 vector, and utþk ¼ ðutþk1 …utþkn Þ.

pffiffiffiffi
From standard GMM, T ð^ ÞNð0,Þ, a
with ¼ Z01 S0 Z01 , Z0 ¼ ðIn Eðxt xt ÞÞ,
and

S0 ¼ EðhtþkhtþkÞ ¼ E ðutþkutþkÞ ðxt x t Þ : ðB3Þ
^ b of S0 is given by
The Hodrick (1992) estimate S T
1
S^Tb ¼ W W , ðB4Þ
T
where W is a T Kn matrix, W ¼ ðWk1 …Wkn Þ, and Wk , T n, is given by

Wk ¼ ðw 1þk ,…w Tþk Þ, and wtþk , K 1, is
!
X
k1
wtþk ¼ etþ1 xti , ðB5Þ
i¼0
since under the null of no predictability the one-step ahead errors etþi ¼ utþ1 are uncorre-
lated and utþk ¼ etþ1 þ … þ etþk . Denoting X ¼ ðx 1 ,…x T Þ, T K, an estimate of Z01 is
given by
1
Z^T1 ¼ ½In ðXXÞ1 : ðB6Þ
T
To test the hypothesis C ¼ 0, we use the Newey (1985) 2 test:
ðC Þ½C ^ 1 C ^ 2
^ C ðB7Þ
rankðCÞ ,
^ ¼Z
with ^b Z
^ 1 S ^ 1
T T T .
698
Appendix C: Testing Predictability Pooling Cross-Sectional Information
C.1 Generalizing Hodrick (1992) to Cross-sectional regressions

To pool cross-sectional country information, we estimate the system
y~itþk ¼ i þ i zit þ uitþk ðC1Þ
for i ¼ 1…N countries, subject to the restriction i ¼ 8i, but impose no restrictions on i
across countries. We take i ¼ United States, United Kingdom, France, and Germany.
Let the dimension of zt be ðK 1Þ so that there will be a total of K regressors, including
the constant terms i for each of N countries. In Equation (C1), we denote the free
parameters ¼ ð1 …N Þ and the unrestricted parameters stacked by each equation
¼ ð1 1 …N N Þ. We can estimate the system in Equation (C1) subject to the restriction
that C ¼ 0, where C is a NK ðN 1ÞðK 1Þ matrix of the form:
0 1
~0 I ~0 I ~0 : : :
B ~0 O ~0 I ~0 I ::: C
B C
C¼B .. C, ðC2Þ
@ . A
~0 O ~0 ~0 I
where ~
0 is a ðK 1Þ 1 vector of zeros, O is a ðK 1Þ ðK 1Þ matrix of zeros, and I is a
ðK 1Þ rank identity matrix.
Denote
y1tþk …~
y~tþk ¼ð~ yN
tþk Þ ðN 1Þ
xit ¼ð1zit Þ ðK 1Þ
utþk ¼ðu1tþk …uN tþk Þ ðN 1Þ
0 1 1 ðC3Þ
xt 0
B .. C
Xt ¼B@ .
C
A ðNK NÞ:
0 xN
t
Then, the system can be written as
y~tþk ¼ Xt þ utþk , ðC4Þ
subject to C ¼ 0. Let Y ¼ ð~y 1þk …~yTþk

Þ, X ¼ ðX 1 …XT Þ, and U ¼ ðu 1 þk …u Tþk
Þ. Then,
the compact system can be written as
Y ¼ X þ U, subject to C ¼ 0: ðC5Þ
A consistent estimate ^ of is given by
699
^ ¼ ols ðXXÞ1 C½CðXXÞ1 C 1 Cols , ðC6Þ
with ols ¼ ðX X Þ1 X Y . This gives us a consistent estimate ^ of .

The moment conditions of the system in Equation (C4) are
Eðhtþk Þ ¼ EðXt utþk Þ ¼ 0:
By standard GMM, ^ has distribution
pffiffiffiffi a
T ð^ Þ Nð0,ðD 0 S01 D0 Þ1 Þ, ðC7Þ
with
@htþk
D 0 ¼ E
@
and
S0 ¼ Eðhtþk h tþk Þ:
^ b of S0 is given by
The Hodrick (1992) estimate S T
1X T
S^Tb ¼ wkt wkt , ðC8Þ
T t¼k
where wkt ðNK 1Þ is given by

!
X
k1
wkt ¼ Xti etþ1 :
i¼0
Under the null hypothesis of no predictability, utþk ¼ etþ1 þ …etþk , where etþ1 are the one-
step ahead serially uncorrelated errors. This is the SUR equivalent of the Hodrick (1992)
estimate for univariate OLS regressions.
An estimate D^ T of D0 is given by
XT
^ T ¼ 1
D
@htþk
,
T t¼0 @
and
where ¼ ð1 …N Þ
2 3
1 z1t 0
6 1 z2t 7
@htþk 6
6 ..
7
7
¼6 . 7:
@ 6 7
4 0 1 zN 5
t
z1t z1t z1t z2t z2t z2t ::: zN
t zN
t zt
N
The estimate ^ has the distribution
700
pffiffiffiffi a ^ T ðS^b Þ1 D

T ð^ Þ Nð0,½D ^ T 1 Þ: ðC9Þ
T
There are ðN þ K 1Þ free parameters in with NK moment conditions. This gives

NK ðN þ K 1Þ overidentifying restrictions. The Hansen (1982) 2 J-test of overidenti-
fying restrictions is given by
J ¼ T ðh ðS^Tb Þ1 hÞ

2 ½NK ðN þ K 1Þ, ðC10Þ
with
1X T
h ¼ htþk :
T t¼0
C.2 Generalizing Cochrane-Orcutt to cross-sectional regressions

We start with the one-step ahead predictive regression in Equation (C4) with k ¼ 1, repeated
here for convenience:
y~tþ1 ¼ Xt þ utþ1 , ðC11Þ
subject to C ¼ 0, where C is given by Equation (C2). We assume that the N 1 vector or

errors utþ1 follows the process:
utþ1 ¼ ut þ "tþ1 , ðC12Þ
where "t are IID with Eð"t "t Þ ¼ . The unconditional covariance matrix of ut is given by:
Eðut ut Þ ¼ devech½ðI Þ1 vecðÞ:
We write Equation (C11) in terms of uncorrelated residuals:
y~tþ1 ~
ytþ1 ¼ ðXt Xt1 Þ þ "tþ1 ,
or as
y*tþ1 ¼ ðXt* Þ þ "tþ1 , ðC13Þ
where y*tþ1 ¼ ~
ytþ1 ~yt and Xt* ¼ Xt Xt1 . To construct y*tþ1 and Xt* , we use a con-
sistent estimate, ^ of . Using the estimate ^ of in Equation (C6), we set
,
P
^ ¼ 1=T vt v t1 , where vt are the residuals vt ¼ ~ytþ1 X t ^ that are standardized to

have unit variance.
To compute Cochrane-Orcutt standard errors for ¼ ð1 : : : N Þ in the pooled system
(C11) with matrix-autocorrelated residuals, we use GMM. The set of NK 1 moment
conditions implied by the regression (C13) are
701
Eðhtþ1 Þ ¼ EðXt* "tþ1 Þ ¼ 0:
^ T of S0 ¼ Eðhtþ1 htþ1 Þ to be
We set the estimate S
1X T
S^T ¼ ^
½X * ðy* Xt* Þ:
T t¼1 t tþ1
^ T , of the derivative of the moment conditions with respect to can be

An estimate, D
computed by taking, and summing, appropriate elements of Xt* Xt* .
Appendix D: Proof of Proposition

The present value relation (5) can be written as
Pt X 1
¼ Mtþi ,
Dt i¼1
where
" !#
X
i1 X
i
Mtþi ¼ Et exp tþj þ gdtþj : ðD1Þ
j¼0 j¼1
We show that
Mtþi ¼ expðai þ bi Xt þ ci t Þ,
which then proves relation (9).

The initial conditions are given by
Et ½expð t þ gdtþ1 Þ ¼ expð t ÞEt ½expðe2 Xtþ1 Þ

1 ðD2Þ
¼ exp t þ e2
þ e2 Xt þ e2 e2 ,
2
where e2 ¼ ð0, 1Þ and equating coefficients yields Equation (11).

To prove the recursive relations (10), we use proof by induction. Suppose that
Mtþi ¼ expðai þ bi Xt þ ci t Þ. Then, we can write:
" !#
X
i X
iþ1
Mtþiþ1 ¼ Et exp tþj þ gdtþj
j¼0 j¼1
¼ Et ½expð t þ gdtþ1 Þ expðai þ b i Xtþ1 þ ci tþ1 Þ

ðD3Þ
1
¼ exp ai þ ci þ ðci 1Þ t þ c2i 2 þ ðe2 þ bi þ ci Þ
2

1
þ ðe2 þ bi þ ci ÞXt þ ðe2 þ bi þ ci Þðe2 þ bi þ ci Þ :
2
Collecting terms, we can write:
702
Mtþiþ1 ¼ expðaiþ1 þ biþ1 Xt þ ciþ1 t Þ,
where aiþ1 , biþ1 , and ciþ1 take the form in Equation (10). The sum of exponential affine
functions of the price-dividend ratio means that this model falls under the class of affine
equity pricing models developed by Ang and Liu (2001), Bekaert and Grenadier (2002), and
Bakshi and Chen (2005).
Appendix E: Estimating the Present Value Model

The calibration of the present value model proceeds in two steps. First, we estimate the VAR
parameters ð
, , Þ in Equation (7) of Xt ¼ ðrt gtd Þ. This estimation is complicated by the fact
that in the data we observe dividends summed up over the past year, but we specify the
frequency of our model to be quarterly. Since the VAR specifies the dynamics of short rates
and dividend growth, we use only short rate and dividend growth data to estimate the VAR. In
the second step, we hold the VAR parameters fixed and estimate the parameters of the discount
rate t in Equation (8). In both stages, we use simulated method of moments (Duffie and
Singleton 1993). To calibrate the model, we use US data from January 1952 to December 2001.
We observe dividend growth summed up over the past 12 months, gtd,4 , but the model
requires quarterly dividend growth gtd . By simulating gtd from the VAR, we can construct
gtd,4 using the transformation

Dt þ Dt1 þ Dt2 þ Dt3
gd,4
t ¼ log
Dt1 þ Dt2 þ Dt3 þ Dt4
d d d d d d
!
1 þ egt2 þ eðgt2 þgt1 Þ þ eðgt2 þgt1 þgt Þ
¼ gdt3 þ log :
d d d d d
1 þ egt3 þ eðgt3 þgt2 Þ þ eðgt3 þgt2 þgt1 Þ
d
ðE1Þ
Equation (E1) shows that the relation between quarterly growth rates and growth rates at a
quarterly frequency using dividends summed up over the past year is highly nonlinear. In
particular, the summing of dividends over the past four quarters induces serial correlation up
to three lags, even when gtd is serially uncorrelated.
To estimate the VAR on Xt , we impose a restricted companion form where 12 ¼ 0, so
there is no Granger-causality from dividend growth to interest rates. This assumption is
motivated by an analysis of an unconstrained VAR on ðrt gtd,4 Þ, where we fail to reject the
null that gtd,4 fails to Granger-cause interest rates. Hence, we first estimate an AR(1) on
quarterly short rates and then holding the parameters for rt fixed, we estimate the remaining
parameters in
, , and by using first and second moments of gtd,4 , in addition to the
moments Eðrt gtd,4 Þ, Eðgt4
d,4 d,4
gt Þ, and Eðrt4 gtd,4 Þ. Hence, the system is exactly identified. The
cross-moment lag length is set at four because the first three lags are affected by the
autocorrelation induced by the nonlinear filter for annual dividend growth in Equation
(E1). We compute the Newey–West (1987) weighting matrix with four lags using the data, so
we do not need to iterate on the weighting matrix.
In the second stage, we hold the parameters of the VAR fixed at their estimates in Panel
A. The parameters , , , and 2 are estimated by matching 12 moment conditions: the
first and second moments of excess returns ~yt and dividend yields D4t =Pt in the data, with
d,4
rt1 and gt1 as instruments. Hence, we use the moments
E½qt zt1 ¼ 0,
yt D4t =Pt y~2t ðD4t =Pt Þ2 and zt1 ¼ ð1 rt1 gt1

where qt ¼ ½~ d,4
Þ.
703
The relation between the closed-form quarterly dividend yields dyt (Proposition 4.1) and
the dividend yields in the data dy4t (which use dividends summed over the last 4 quarters) is
complex:
D4t Dt þ Dt1 þ Dt2 þ Dt3

¼
Pt Pt
ðE2Þ
Dt Dt1 Pt1 Dt2 Pt2 Dt3 Pt3
¼ þ þ þ ,
Pt Pt1 Pt Pt2 Pt Pt3 Pt
where the capital gain over n periods Pt =Ptn can be evaluated using:
!
Pt Pt =Dt X
n1
¼ exp gdti :
Ptn Ptn =Dtn i¼0
The predictability regressions use excess log returns y~tþ1 ¼ logðYtþ1 Þ rt , where
Yt ¼ logððPtþ1 þ Dtþ1 Þ=Pt Þ. Since we model t ¼ logðEt ðPtþ1 þ Dtþ1 Þ=Pt ÞÞ, the expected
log excess return Et ðlogðYtþ1 Þ rt Þ 6¼ ðlog Et ðYtþ1 Þ rt Þ is not closed form in our model
but is a function of time t state variables because of the Markov structure of the model. The
two quantities Et ðlogðYtþ1 Þ rt Þ and ðlog Et ðYtþ1 Þ rt Þ are not equal because of the pre-
sence of state-dependent heteroskedasticity, which induces (time-varying) Jensen’s inequal-
ity terms. To compute the conditional expected value of k-period excess returns, we project
excess returns onto a fourth-order polynomial in ðXt , t Þ, and the fitted value is the model-
implied conditional expected excess return.
References
Amihud, Y., and C. M. Hurvich, 2004, ‘‘Predictive Regressions: A Reduced-Bias Estimation Method,’’
Journal of Financial and Quantitative Analysis, 39(4), 813–841.
Ang, A., and G. Bekaert, 2001, ‘‘Stock Return Predictability: Is It There?’’ NBER working paper 8207.
Ang, A., and G. Bekaert, 2002, ‘‘Regime Switches in Interest Rates,’’ Journal of Business and Economic
Statistics, 20(2), 163–182.
Ang, A., and J. Liu, 2001, ‘‘A General Affine Earnings Valuation Model,’’ Review of Accounting Studies,
6, 397–425.
Ang, A., and J. Liu, 2006, ‘‘Risk, Return and Dividends,’’ Forthcoming in Journal of Financial Economics.
Bakshi, G., and Z. Chen, 2005, ‘‘Stock Valuation in Dynamic Economies,’’ Journal of Financial Markets,
8(2), 115–151.
Bansal, R., and C. Lundblad, 2002, ‘‘Market Efficiency, Fundamental Values and Asset Returns in
Global Equity Markets,’’ Journal of Econometrics, 109, 195–237.
Bansal, R., and A. Yaron, 2004, ‘‘Risks for the Long Run: A Potential Resolution of Asset Pricing
Puzzles,’’ Journal of Finance, 59(4), 1481–1509.
Bekaert, G., and S. Grenadier, 2002, ‘‘Stock and Bond Pricing in an Affine Economy,’’ Working Paper,
Columbia Business School, New York.
Bollerslev, T., and R. J. Hodrick, 1996, ‘‘Financial Market Efficiency Tests,’’ in M. H. Pesaran and M. R.
Wickens (eds), The Handbook of Applied Econometrics Vol 1, Macroeconomics, Blackwell, Oxford, United
kingdom, 415–458.
Bossaerts, P., and P. Hillion, 1999, ‘‘Implementing Statistical Criteria to Select Return Forecasting
Models: What do we Learn?,’’ Review of Financial Studies, 12, 405–428.
704
Boudoukh, J., and M. Richardson, 1993, ‘‘The Statistics of Long-Horizon Regressions,’’ Mathematical
Finance, 4(2), 103–120.
Breen, W., L. R. Glosten, and R. Jagannathan, 1989, ‘‘Economic Significance of Predictable Variations in
Stock Index Returns,’’ Journal of Finance, 44(5), 1177–1189.
Campbell, J. Y., 1987, ‘‘Stock Returns and the Term Structure,’’ Journal of Financial Economics, 18,
373–399.
Campbell, J. Y., 1991, ‘‘A Variance Decomposition for Stock Returns,’’ Economic Journal, 101, 157–179.
Campbell, J. Y., 2003, ‘‘Consumption-Based Asset Pricing,’’ in G. Constantinides, M. Harris, and

R. Stulz (eds), Handbook of the Economics of Finance, North-Holland, Amsterdam, 803–887.
Campbell, J. Y., and R. J. Shiller, 1988a, ‘‘Stock Prices, Earnings and Expected Dividends,’’ Journal of
Finance, 43(3), 661–676.
Campbell, J. Y., and R. J. Shiller, 1988b, ‘‘The Dividend-Price Ratio and Expectations of Future
Dividends and Discount Factors,’’ Review of Financial Studies, 1(3), 195–228.
Campbell, J. Y., and L. M. Viceira, 1999, ‘‘Consumption and Portfolio Decisions When Expected
Returns are Time Varying,’’ Quarterly Journal of Economics, 114(2), 433–492.
Campbell, J. Y., and M. Yogo, 2006, ‘‘Efficient Tests of Stock Return Predictability,’’ forthcoming in
Journal of Financial Economics, 81(1), 27–60.
Clarida, R., J. Galı́, and M. Gertler, 1999, ‘‘The Science Of Monetary Policy: A New Keynesian
Perspective,’’ Journal of Economic Literature, 37(4), 1661–1707.
Cochrane, J. H., 1992, ‘‘Explaining the Variance of Price-Dividend Ratios,’’ Review of Financial Studies,
5, 243–280.
Cochrane, J. H., 2001, Asset Pricing, Princeton University Press, Princeton NJ.
Duffie, D., and K. J. Singleton, 1993, ‘‘Simulated Moments Estimation of Markov Models of Asset
Prices,’’ Econometrica, 61(4), 929–952.
Elliot, G., and J. H. Stock, 1994, ‘‘Inference in Time Series Regression when the Order of Integration of a
Regressor is Unknown,’’ Econometric Theory, 10, 672–700.
Engstrom, E., 2003, ‘‘The Conditional Relationship between Stock Market Returns and the Dividend
Price Ratio,’’ Working Paper, Columbia University, New York.
Fama, E., and F. French, 1988, ‘‘Dividend Yields and Expected Stock Returns,’’ Journal of Financial
Economics, 22, 3–26.
Fama, E., and G. W. Schwert, 1977, ‘‘Asset Returns and Inflation,’’ Journal of Financial Economics, 5, 115–146.
Ferson, W. E., S. Sarkissian, and T. T. Simin, 2003, ‘‘Spurious Regressions in Financial Economics?’’
Journal of Finance, 58(4), 1393–1414.
Goetzmann, W. N., and P. Jorion, 1993, ‘‘Testing the Predictive Power of Dividend Yields,’’ Journal of
Finance, 48(2), 663–679.
Goetzmann, W. N., and P. Jorion, 1995, ‘‘A Longer Look at Dividend Yields,’’ Journal of Business, 68,
483–508.
Goyal, A., and I. Welch, 2003, ‘‘The Myth of Predictability: Does the Dividend Yield Forecast the Equity
Premium?’’ Management Science, 49(5), 639–654.
Goyal, A., and I. Welch, 2004, ‘‘A Comprehensive Look at The Empirical Performance of Equity
Premium Prediction,’’ NBER Working Paper 10483.
Hansen, L., 1982, ‘‘Large Sample Properties of Generalized Method of Moments Estimators,’’ Econo-
metrica, 50, 1029–54.
705
Hansen, L., and R. Hodrick, 1980, ‘‘Forward Exchange Rates as Optimal Predictors of Future Spot
Rates: An Econometric Analysis,’’ Journal of Political Economy, 88, 829–853.
Hodrick, R. J., 1992, ‘‘Dividend Yields and Expected Stock Returns: Alternative Procedures for Inference
and Measurement,’’ Review of Financial Studies, 5(3), 357–386.
Jansson, M., and M. J. Moreira, 2006, ‘‘Optimal Inference in Regression Models with Nearly Integrated
Regressors,’’ forthcoming in Econometrica.
Kwiatkowski, D., P. C. B. Phillips, P. Schmidt, and Y. Shin, 1992, ‘‘Testing the Null Hypothesis of
Stationarity Against the Alternative of a Unit Root: How Sure are we that Economic Time Series have a
Unit Root?’’ Journal of Econometrics, 54(1–3), 159–178.
Lamont, O., 1998, ‘‘Earnings and Expected Returns,’’ Journal of Finance, 53(5), 1563–1587.
Lee, B. S., 1992, ‘‘Causal Relations Among Stock Returns, Interest Rates, Real Activity and Inflation,’’
Journal of Finance, 48, 1591–1603.
Lettau, M., and S. Ludvigson, 2001, ‘‘Consumption, Aggregate Wealth and Expected Stock Returns,’’
Journal of Finance, 56(3), 815–849.
Lettau, M., and S. Ludvigson, 2005, ‘‘Expected Returns and Expected Dividend Growth,’’ Journal of
Financial Economics, 76, 583–626.
Lewellen, J., 2004, ‘‘Predicting Returns with Financial Ratios,’’ Journal of Financial Economics, 74,
209–235.
LO, A W., and A. C., Mackinlay ‘‘Data-Snooping Biases in Tests of Financial Asset Pricing Models’’
Review of Financial Studies, 3(3), 431–467.
Menzly, L., J. Santos, and P. Veronesi, 2004, ‘‘Understanding Predictability,’’ Journal of Political
Economy, 112(1), 1–47.
Newey, W., 1985, ‘‘Generalized Method of Moments Specification Testing,’’ Journal of Econometrics,
29(3), 229–256.
Newey, W., and K. West, 1987, ‘‘A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrela-
tion Consistent Covariance Matrix,’’ Econometrica, 55, 703–708.
Patelis, A. D., 1997, ‘‘Stock Return Predictability and the Role of Monetary Policy,’’ Journal of Finance,
52(5), 1951–1972.
Phillips, P. C. B., and P. Perron, 1988, ‘‘Testing for a Unit Root in Time Series Regression,’’ Biometrica,
75, 335–346.
Polk, C., S. Thompson, and T. Vuolteenaho, 2006, ‘‘New Forecasts of the Equity Premium,’’ Journal of
Financial Economics, 81(1), 101–141.
Richardson, M., 1993, ‘‘Temporary Components of Stock Prices: A Skeptic’s View,’’ Journal of Business
and Economic Statistics, 11(2), 199–207.
Richardson, M., and T. Smith, 1991, ‘‘Tests of Financial Models in the Presence of Overlapping
Observations,’’ Review of Financial Studies, 4(2), 227–254.
Richardson, M., and J. H. Stock, 1989, ‘‘Drawing Inferences from Statistics Based on Multiyear Asset
Returns,’’ Journal of Financial Economics, 25, 323–348.
Shiller, R. J., and A. E. Beltratti, 1992, ‘‘Stock Prices and Bond Yields: Can Their Comovements be
Explained in Terms of Present Value Models?’’ Journal of Monetary Economics, 30, 25–46.
Stambaugh, R. F., 1999, ‘‘Predictive Regressions,’’ Journal of Financial Economics, 54, 375–421.
Timmermann, A, and B. Paye, 2006, ‘‘Instability of Return Prediction Models,’’ Journal of Empirical
Finance, 13, 274–315.
706
Torous, W., R. Valkanov, and S. Yan, 2004, ‘‘On Predicting Stock Returns with Nearly Integrated
Explanatory Variables,’’ Journal of Business, 77(4), 937–966.
Valkanov, R., 2003, ‘‘Long-Horizon Regressions: Theoretical Results and Applications,’’ Journal of
Financial Economics, 68(2), 201–232.
Wolf, M., 2000, ‘‘Stock Returns And Dividend Yields Revisited: A New Way To Look At An Old
Problem,’’ Journal of Business and Economic Statistics, 18(1), 18–30.
707

How Financial Motivation Can Be Measured in A Regression Model?

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

How Financial Motivation Can Be Measured in A Regression Model?

Загружено:

Авторское право:

Доступные форматы

Stock Return Predictability: Is it There?

In a rational no-bubble model, the price-dividend ratio is the expected

Our main findings can be summarized as follows. First, the statistical

regression provides a rather poor proxy to true expected returns. How-

where D4t ¼ Dt þ Dt1 þ Dt2 þ Dt3 represents dividends summed over

1.1 Long sample data

Excess Short Dividend Earnings Dividend Earnings

Panel A: Long-sample data

Excess Short Dividend Earnings Dividend Earnings

Correlations of excess returns, June 1953–December 2001

Panel B: MSCI data

Correlations of excess returns

United States, United Kingdom, and Germany. For the United

interpreting the evidence on excess return predictability using the divi-

1.2 Short sample MSCI data

2. The Predictability of Equity Returns

2.1 Predictability regressions

ytþk ¼ k þ k zt þ "tþk,k ,

where ~ ytþk ¼ ð=kÞ½ðytþ1 rt Þ þ ::: þ ðytþk rtþk1 Þ is the annualized

We show in section 5 that the performance of Hodrick (1992) standard

2.2 Return predictability in the United States

1935-2001 Dividend Yield Coefficient 1935-2001 T-statistic

1952-2001 Dividend Yield Coefficient 1952-2001 T-statistic

1935-1990 Dividend Yield Coefficient 1935-1990 T-statistic

Newey–West errors, we use k þ 1 lags. The coefficient pattern is similar

Panel A: quarterly S&P data

k-mths dy12 r dy12 2 test

Panel B: monthly MSCI data

We estimate regressions of the form ~ytþk ¼ k þ zt þ tþk,k where y

2.3 Predictability of excess returns in four countries

k-qtrs dy4 only J-test r dy4 2 Test J-test

Panel A: quarterly long-sample data

k-mths dy12 only J-test r dy12 chi 2 test J-test

Panel B: monthly MSCI data

Germany and around 1:5 for France. The coefficient monotonically

3. Do Dividend Yields Predict Cash flows or Interest Rates?

expð t Þ ¼ Et ½ðPtþ1 þ Dtþ1 Þ=Pt  Et ½Ytþ1 : ð4Þ

assuming a transversality condition holds. Note that Equation (5) is

where c and are linearization constants. Equation (5) is an exact

3.1 Dividend growth predictability

k-qtrs dy4 only r dy4 2 test

Panel A: US S&P data

k-mths dy12 only r dy12 2 test

horizons. The magnitude of the coefficients is preserved in the bivariate

3.2 Interest Rate Predictability

k-qtrs dy4 Detrended dy4

Panel B: pooled United States, United Kingdom and Germany

k-mths dy12 Detrended dy12

Panel C: pooled United States, United Kingdom, Germany, and France

4. A Present Value Model for Stock Returns

4.1 The model

Proof: See Appendix D.

characterizes rational pricing or only develop approximate solutions. In

Another important special case of the model is constant expected excess

Panel A of Table 6 reports the VAR parameters. Dividend growth

Panel A: estimates of the VAR for ðrt gtd Þ

Null 1 Null 2 Alt 1 Alt 2 Alt 3

Panel B: estimates of the discount rate process

excess returns by setting r ¼ 1 and gd ¼ ¼ ¼ 0. The first model under

4.3 Economic implications

4.3.1 Moments and price-dividend ratio variance decomposition. Table 7

where varðP=D4 Þ is the variance of price-dividend ratios implied by the

ytþk ¼ k þ k zt þ "tþk,k ,

where ~ ytþk ¼ ð=kÞ½ðytþ1 rt Þ þ ::: þ ðytþk rtþk1 Þ is the annualized

k-mths dy12 r dy12 2 test

We estimate regressions of the form ~ytþk ¼ k þ zt þ tþk,k where y

k-qtrs dy4 only J-test r dy4 2 Test J-test

k-mths dy12 only J-test r dy12 chi 2 test J-test

expð t Þ ¼ Et ½ðPtþ1 þ Dtþ1 Þ=Pt Et ½Ytþ1 : ð4Þ

k-qtrs dy4 only r dy4 2 test

k-mths dy12 only r dy12 2 test

Panel A: estimates of the VAR for ðrt gtd Þ

Univariate Bivariate 2 tests

Univariate regression Lamont regression Trivariate regression 2 tests

Univariate regression Lamont regression Trivariate regression 2 tests

Univariate regression Lamont regression Trivariate regression 2 tests

y~tþk1 ¼ k1 þ k1 zt þ utþk1

We produce an estimate ^ of ¼ ðk1 k1 …kn kn Þ by performing OLS on each equation.

where xt ¼ ð1 z t Þ, a K 1 vector, and utþk ¼ ðutþk1 …utþkn Þ.

To test the hypothesis C ¼ 0, we use the Newey (1985) 2 test:

y~itþk ¼ i þ i zit þ uitþk ðC1Þ

y~tþk ¼ Xt þ utþk , ðC4Þ

subject to C ¼ 0. Let Y ¼ ð~y 1þk …~yTþk

^ ¼ ols ðXXÞ1 C½CðXXÞ1 C 1 Cols , ðC6Þ

with ols ¼ ðX X Þ1 X Y . This gives us a consistent estimate ^ of .

By standard GMM, ^ has distribution